US 5,537,590 AGrant
Apparatus for Applying Analysis Rules to Data Sets in a Relational Database to Generate a Database of Diagnostic Records Linked to the Data Sets
Issue Date:1996-07-16
•8 Claims
•43 Drawing Sheets
Abstract
This invention addresses the need to map information from databases and reports to a new dimension of structured, intelligent interpretations or diagnostics of that information, and then querying in a coordinated manner both the original information and the resultant diagnostics databases. The invention combines elements currently available in decision support tools, programming languages and expert system building tools. First, querying mechanisms for the information in reports and databases is provided. Second, both a test processing engine and an Expert System run a set of if-then-else tests and expert rules on said information, and the resulting coded diagnostics are stored in a diagnostics database. In a simple implementation, first, the information database contains a single flat database file, second each if-then-else test compares particular data items in this file against other data items or against absolute values and, if true, generates one or more data pairs containing the test identifier and each one of the data item identifiers involved in the test. The invention is completed, third, by a querying engine and a EIS-like reporting system, both capable of structuring, filtering, linking and querying in a coordinated manner both the original information and the resultant diagnostics databases, and capable of building reports.
Metadata
Inventor
- Armando Amado
Application Information
Application Number:US 1025813
Filing Date:1993-08-05
Priority Date:1993-08-05
Art Unit:237
Classifications
IPC:
G06F 1700G06F 1730
Field of Search:
395371600;575;50;51;52;55;911;153;15715.1;29.1;19;21.1
Patent Drawings (43 sheets)
Description
BACKGROUND OF THE INVENTION
The invention as a whole relates to the fields of decision-support software and expert systems.
More specifically, the invention relates to computer software integrating the functions of expert systems and database query systems for executive report analysis.
Current software implementations do not address directly the need for a system integrating (a) report analysis via the generation of diagnostics using an expert system, (b) storage of said diagnostics in a database related to the original information in the reports, and (c) a query system, empowering the user to search for any information in both reports and diagnostics in a coordinated manner.
Heretofore, no instruments have been involved in a similar idea as that dealt with in the invention. However, its individual elements respond to prior art in the following areas: (A) decision-support software and executive information systems, and (B) expert systems, expert system building tools and other artificial intelligence technologies such as neural networks. Other prior art of lesser relevance to this invention includes: (C) programming tools and languages, (D) relational database systems and (E) management information systems.
A. Referring to DECISION-SUPPORT SOFTWARE, existing tools do not address the specific functionality of this invention.
Decision-support software and executive information systems (EISs) have been primarily designed for business managers. These systems allow managers to see very simple summary reports from huge amounts of information stored in databases. Simplicity is foremost in these systems, since managers don't want to deal with menus, and even less with any programming. So, new concepts such as "downdrilling" are introduced. Typically, the first screen the manager will see is a very summarized report. This report could show a few numbers. The manager will want to know more about some number in the screen, so he simply points at that particular number and he will get a more detailed report associated to it. And in this detailed report, he can again do the same, and get another more detailed and specific report. This is called "downdrilling." EISs also provide a programming language and a set of tools for developing simple reports. In other words, EIS systems are for programmers to build simple reporting applications for managers. And then the managers run the EISs with the applications that have already been developed for them.
A few tools will be now discussed as good examples of the current state of prior art.
The DECISION PAD (V.2.0).TM. decision-support tool, by Apian Software of Menlo Park, Calif., combines facts and opinions in spreadsheet-style matrix. Applications include purchasing, employee evaluations, hiring, vendor evaluations, investing, project prioritization, and sales. The product handles up to 250 Alternatives, 250 Criteria, and 60 Evaluators. Weights are by numbers in 1 or 2 levels or by words. Pre-formatted reports include worksheet, bar graphs, sensitivity analysis, scatter plots and logged what-if cases. The product imports and exports files.
The MAXCIM DECISION SUPPORT SYSTEMS II.TM. decision-support tool, by ASK Computer Systems, Inc. of Mountain View, Calif., provides budget upload function supporting file updates, multiple spreadsheet formats, multiple budgets, selective period updating, quarterly spreading and audit trail. The product includes financial report writer with column and row report definitions, user definable formats, financial management reporting and GL integration.
The DECISIONMAKER.TM. decision-support tool, by ASK Computer Systems, Inc. of Mountain View, Calif., gathers information and presents it on screen. The product provides exception reporting; summary information for manufacturing, marketing and financial areas of company; and supporting detail information for specific inquiries.
The EIS.TM. (Executive Information Services.TM.) decision-support tool, by Boeing Computer Services of Seattle, Wash., solves business planning and control problems. Its features include consolidation, modeling, graphics, reports, statistics and financial functions, backward iteration and sensitivity analysis. It is capable of solving simultaneous equations, recording equations, forecasting, estimating, planning, budgeting and performance tracking. It belongs to the class of systems called `authoring systems`.
The Business Insight.TM. decision-support tool, by Business Resource Software of Austin, Tex., is a powerful tool for business analysis, strategic planning, and forecasting. It is capable of providing expert help in the development of promotional strategies, sales methods and inventory management applications. It is related to financial planning systems, spreadsheets and expert systems.
The Decision Support System.TM. (DSS.TM.) decision-support tool, by Definitive Software Inc. of Bloomington, Ind. is an end-user report writing tool. It provides access to computer files and writes reports using menus.
The EIS Tool Kit.TM. decision-support tool, by Executive Performance Systems of Glendale, Ariz., is a complete executive information system (EIS) which includes all programs and drivers necessary to construct complete and functional EIS applications. It enables user to select input and output devices, arrange hierarchy of data to be displayed and customize display screens.
The EIS Toolkit.TM. decision-support tool, by Ferox Microsystems, Inc. of Alexandria, Va., allows the user to build a financial executive information system using his own account structure. This product provides EIS capabilities such as drill-down from summary detail, color exception reporting, trend analysis, graphics and textual annotation of numeric data.
The Executive Information System.TM. decision-support tool, by Global Software, Inc. of Raleigh, N.C., provides icon-based, graphical views of summarized information. It features host logon and delivery of data through scheduled, unattended batch jobs to individual stations. It generates reports through the use of a mouse or with touch screen drivers.
The Forest & Trees.TM. decision-support tool, by Channel Computing Inc. of Newmarket, N.H., is a data access software tool that monitors vital signs of small to large business. It works in the background alerting the user of any unusual data. It provides access and monitoring capabilities to databases and spreadsheets on individual PCs, networked PCs and minicomputers. It provides drill-down capabilities that allow users to access supporting data. It restructures data into visual set of business vital signs that summarize enterprise's status and health. It includes graphics capabilities.
The LightShip.TM. decision-support tool, by Pilot Executive Software, of Boston, Mass., is another popular authoring system for the development of executive information systems.
Of all other executive information systems examined, none provides the specific functionality of this invention. Many of these EIS tools are simply multidimensional spreadsheets, interesting for managers because of the flexibility to recalculate and examine information in different manners and dimensions. A good example of these systems is the CA-COMPUTE.TM. multi-dimensional spreadsheet program, by Computer Associates Corporation of Islandia, N.Y.
B. Referring to EXPERT SYSTEMS and expert systems building tools, it is important to state that Expert systems, commonly known as knowledge systems, have been widely known for at least ten years. However, their technology is an important prior art reference to this invention, so it will be briefly described here.
Researchers define expert systems (also called knowledge systems) in the following way: Intelligent computer programs that use knowledge and inference procedures to solve problems that are hard enough as to require in their solution, significant expertise.
Expert systems typically consist of (a) an interpretive language where the user may write his or her program statements and the conditions associated with those statements, (b) an inference engine, which provides the mechanism through which the expert rules are interpreted and fired, and (c) an executive front-end or expert shell, that helps users write application programs using the language, and that helps them run the expert applications developed, and that also helps them develop and query reports or the generated diagnostics.
With expert systems, in contrast than with other types of programs, users tell the computer what to know, not what to do. To build a traditional program, any developer creates a set of instructions. To build an expert system, the developer creates knowledge. Traditional programs execute by following every step of the algorithms contained in the program. Expert systems execute in different ways.
If an algorithm has to execute step by step to find a solution, that algorithm is really a traditional program. If there isn't a step-by-step method available to solve a particular problem, then artificial intelligence techniques such as expert systems (knowledge systems) and neural networks must be used.
Expert (knowledge) systems contain two basic elements: inference engine and knowledge base. The knowledge base holds all information related to the tasks at hand: (a) the rules and (b) the data on which they will be applied. The inference engine is a mechanism that can operate the information contained in the knowledge base.
In a rule-based system, the knowledge base is divided into a set of rules and working memory (or database).
Just like an IF-THEN sentence, each rule has two parts: a premise and a conclusion. A rule is said to be fired when the inference engine finds the premise is stored as TRUE in working memory (the knowledge base) and it incorporates the conclusion of the rule to the working memory (knowledge base) too.
Working memory is the database contained in the knowledge base. This holds all facts that describe the current situation. Generally, the expert system will start with very few facts. These will expand as the system learns more about the situation at hand, and as far as some rules are executed.
The inference engine or rule interpreter has two tasks. First, it examines facts in working memory and rules in the rule base, and adds new facts to the database (memory) when possible. That is, it fires rules. Second, it determines in what order rules are scanned and fired.
The inference engine can determine the order in which rules should be fired by different methods such as forward chaining, backward chaining, breadth- or depth-wise scan techniques, etc. Applications that use forward chaining, such as process control are called data-driven. Applications that use backward chaining are called goal-driven. Forward chaining should be used for small sets of relevant facts, where many facts lead to few conclusions. A forward chaining system must have all its data at the start, rather than asking the user for information as it goes. Backward chaining should be used for applications having a large set of facts, where one fact can lead to many conclusions. A backward-chaining system will ask for more information if needed to establish a goal.
Some systems' inference engines are designed to ask the user for more information, and to inform the user about the conclusions that have been reached. After the inference engine shows a conclusion to the user, the user may ask "how" and the system will explain how that given value was found. After the inference engine asks a question to the user, the user may also ask "why" and the system will explain why the current question is being asked.
Development of serious expert system implementations is a non-trivial task, usually left to a programmer experienced in the art. Typically, expert system development kits provide a programming language for the development of specific expert applications. A separate element in expert system building tools, commonly known in the industry as an expert shell, also provides the programmer and the user with a front-end application designed to ease the design-program-run cycle in the development and use of an expert system application.
Many expert system building tools are available to the public, such as the VP-EXPERT.TM. expert shell by Paperback Software, the CLIPS.TM. expert system development tool and language by NASA, Nexpert Object.TM. expert system development tools by Neuron Data, and the KnowledgeMaker.TM. development system by Knowledge Garden. On their part, NEURAL NETWORKS attempt to mimic the human brain by "learning" different sets of stimulus patterns (such as medical symptoms) and their associated responses (diagnoses). Incomplete and/or overlapping sets of stimuli can be presented to the neural network, which can then return several responses matching those stimuli using probability weightings to produce an ordered list of responses.
Each neural network problem session contains a set of defined stimuli, a set of defined responses, and a set of relationships between specific groups of stimuli and the response that each group is to produce. The set of stimuli (responses) is represented by a group of stimulus (response) nodes at what is called the "input (output) layer". Then, there is usually one or more intermediate layers, containing nodes that are each linked to every input layer node and every output layer node in the network. The number of the middle layer nodes is usually equal to the average of the number of input and output nodes. Probability values (weights) are then associated with each of these connections and are constantly being updated as the network "learns" new information.
None of the expert systems, expert system building tools or neural network building tools in the prior art are capable of automatically creating an integral database of diagnostics that can be queried simultaneously and in a synchronized manner with the original data. A few existing systems would allow the replication of the operations of this invention using their general-purpose programming languages, but it would be very complicated to program to do the functions equivalent to the claimed invention. To do such a thing would require a fair degree of sophistication in the user, and to do what the claimed invention does is not at all suggested by any prior art expert system or expert system tool, since their teachings are very general in this respect. Only with the benefit of hindsight and a great deal of programming skill would one skilled in the art appreciate that what is done by the claimed invention could be done with the prior art tools and be able to do it. The claimed invention is not obvious from any prior art, because prior art products, tools and theories do not mention building a diagnostics data structure and linking of this structure with the original data along the lines of the claimed invention.
The same could be said of existing languages that are particularly oriented to the development of expert systems other artificial intelligence products and technologies, such as LISP, PROLOG and others.
C. Referring to PROGRAMMING TOOLS AND LANGUAGES, technology and prior art allow the user to easily integrate the operation of independent programs running concurrently, by defining how specific predetermined data should be shared. Good examples of this technology comprise the MICROSOFT WINDOWS.TM. graphical operating environment, by Microsoft Corporation, the NEWWAVE.TM. front-end tool, by Hewlett Packard corporation, the X-WINDOWS.TM. graphical operating environment and the X-MOTIF.TM. graphical operating environment. However, these tools only provide a very general framework where all the programming would still have to be done, if possible, to create a logic procedure equivalent to that of this invention. Furthermore, given the current state of technology and the prior art, it was considered much more practical and easier to develop the best-mode implementation of this invention using from a general-purpose programming language.
D. Referring to relational database systems, a wide number of these systems exist in the market today, but all of these should be interpreted as general-purpose programming languages for the operation of information stored in the relational database format. None of these systems can replicate the operation of this invention without a sizable programming investment.
E. Referring to management information systems (MIS), most current implementations of these systems do not incorporate decision-support tools, expert systems, or an integrated system of generating and querying reports and their associated diagnostics. When a MIS incorporates a decision-support tool, this tool should be interpreted as a separate instrument. MIS systems may integrate data from a whole distributed organization, but nothing has been suggested in the prior art about the generation of a diagnostics database linked to the original data.
As it has already been said, the prior art tools described in sections A to E would be very complicated to program to do a function equivalent to the claimed invention. To do such a thing would require a fair degree of sophistication in the user, and to do what the claimed invention does is not at all suggested by these tools in the prior art. The teachings of the prior art's instruments and programming tools are very general. Only with the benefit of hindsight and a great deal of programming skill would one skilled in the art appreciate that what is done by the claimed invention could be done with the existing programming tools.
The cited PRIOR-ART references are important because the invention provides the benefits of integrating their main characteristics. (A) Like decision--support software and executive information systems, the invention allows managers to see very simple summary reports from huge amounts of information stored in databases. Simplicity is foremost in the invention, as it has been proved in the preferred embodiment implementation. Typical EIS tools such as downdrilling are also used in the invention Typically, the first screen the manager will see is a very summarized report, then the user can get more and more detailed information on those choices he makes. The invention can alert the user of any unusual data. Only very few EIS systems can do this, and they do it only by coloring the data or sounding audible alarms in the computer while the invention builds a whole structured database of diagnostics. Like EIS systems, the invention is capable of restructuring data into visual set of business vital signs that summarize any enterprise's status and health, but the invention will also be able to do this structuring of the information, based on the associated diagnostics' structure. No other system can do this.
(B) Like expert system building tools, the invention allows the definition of expert rules and clauses, and data can be integrated and processed through an inference engine, presenting the final results of the process to the user. Like expert systems, neural networks and other artificial intelligence techniques, the invention is capable of finding solutions to expert analysis problems, summarizing a large body of information into a compact, structured knowledge database.
(C) Like programming tools and languages, the invention allows flexible development of applications in many fields of human knowledge. The expert system--inference engine module in the invention comprises, in fact, a full development programming language.
(D) Like relational database systems, the invention allows data structuring of its basic input information and diagnostic results: relational database indexing, filtering, and data query operations of all types. Full applications can be integrated with the invention through indexing operations between their key files.
(E) Finally, like management information systems, the invention can integrate information of all kinds in a unified environment. The invention is capable of operating this information and presenting appropriate results to managers.
The invention adds a scheme for the integration of data handling benefits and querying techniques from relational database systems, flexible development of applications from programming tools, integration of information from MIS tools, expert analysis from expert systems, and decision support and reporting capabilities from decision support tools. This form of integration that the claimed invention does is not at all suggested by these tools in the prior art.
OBJECTS AND ADVANTAGES OF THE INVENTION
Accordingly, the following paragraphs describe several OBJECTS AND ADVANTAGES of the invention.
The invention is useful for the analysis of reports and structured data in almost all types of information and applications. The invention can process and generate in just seconds large structured databases of diagnostics to be queried immediately. The analysis of these same reports using any prior art product or concept may require a valuable time investment of many hours on the part of the user.
The invention lets managers concentrate in action and strategy, that is, in their final goals and results. The invention executes all repetitive report analysis, ratios and number comparisons, signals for dangers, menaces, strengths and opportunities. While all other products are only able of showing averages of all available information, the invention is capable of showing either database summaries or individual case analysis classified by degree of importance and relevance to the management process. And all these concepts can respond to state-of-the-art artificial intelligence, neural networks, and expert system analysis techniques.
Therefore, the invention represents important time and cost reductions to managers, administrators, technical people, manufacturing engineers, etc.
With state-of-the-art technologies, data integration has meant easy access to all data, and the production of summary reports representing the condition of the organization, firm or institution based on a set of averages. For end-users, all this translates to data accumulation and proliferation. This makes data handling ever more complicated as time goes on. On the other hand, the invention immediately shows priorities according to the managers' own needs, preferences and criteria. The invention could substitute current EIS and data analysis technologies, but it can better serve the needs of end-users by integrating this new technology to existing tools. Thus, the invention can help managers and end-users determine which information and which tool they should check and when they should do so.
As information grows, data classification schemes become more important. The invention allows data organization by function, action and end-user preferences, rather than by type of data.
Consulting expertise, managerial expertise and technical expertise in many fields is often lost as soon as the person responsible leaves the firm and the organization. This is, for example, a common problem in most consulting firms. These firms would like to be able to use a tool that will extend the results of that expertise with their clients. By leaving an embodiment of the invention with a client, programming with the consulting firm's expertise, the consulting firm will be able to extend their presence with the client. Also, the invention can be applied to strategic planning, organizational analysis, decision-analysis, manufacturing control, quality control, and many other fields.
Up to now, artificial intelligence and expert systems have been able to work on a case-by-case strategy. The typical expert system is given one simple problem and it suggest one practical solution to the problem. But these tools are more useful when dealing with vast amounts of data, chaining recommendations and their logical expert inferences in a structured database of diagnostics. The invention allows this.
The invention addresses all information analysis problems, receiving as input all relevant information from structured databases and reports. The invention replicates the typical manager's mechanical ratio analyses on the data, number comparisons and many other operations with the data. The invention may then apply If-then-else clauses, Expert system clauses and other artificial intelligence techniques to the data, such as neural network technologies. All results from these expert data analysis techniques is stored in a relational database, keeping two-directional pointers to and from the particular data the results have been obtained from. The invention finally presents a query system, allowing end users to structure, organize and classify all diagnostics and the original data queries according their own interests, such as people responsible, urgency of relevant matters, actions that should be taken, etc.
Prior art tools such as the Forest & Trees.TM. decision-support tool, by Channel Computing Inc., are capable of differentiating relevant data by using different colors to those numbers that require the manager's attention. In order to do this, these programs allow the definition of If-then-else formulas that can be linked to individual data items in reports and databases. But prior art tools don't allow the creation, storage and coordinated query of a structured database of diagnostics.
The invention is the only tool that shows where is the problem, in a much more efficient and functional way than prior art tools: the invention can present sorted, filtered and structured lists of all diagnostics and their results, according to related actions, persons responsible, priorities, or any other criteria selected by the user.
The invention builds and keeps links between all diagnostics and the original data (reports, typical business databases, etc.). If the end-user queries the invention about a symptom, then the invention will show the related diagnostic or diagnostics, and then the invention can show the original data item or items those diagnostics come from.
The invention's structure and logic allow its integration to most existing applications. In one window, the user will still operate with his favorite application (such as an accounting application). While he moves the cursor over the data in that application, he will be able to see related diagnostics, generated by the invention. If the user switches control to the invention's window, then he will be able to generate, filter, sort and query lists of diagnostics resulting from the data in the linked application. By moving the cursor over each resulting diagnostic, the linked application's data related to the particular diagnostic will be shown in the application's window.
The invention can also be linked in the same way to other typical applications that process or contain data such as statistical analysis applications, spreadsheet or database applications, inventory analysis applications, accounting programs, financial or market analysis programs, etc.
Probably the most powerful feature of the invention is that users will be able to develop all kinds of practical and even self-modifiable strategic analysis applications. Recursive expert analysis techniques can work in the following way: (a) A first pass at the data, using If-then-else formulas will generate a large number of diagnostics. (b) These diagnostics will be stored in a diagnostics database. (c) Now the invention sees these resulting diagnostics as additional data. (d) A second set of formulas is run using these diagnostics as data. (e) More diagnostics are generated and stored in the diagnostics database. (f) Steps `c`, `d` and `e` are repeated over and over until no more diagnostics are generated.
Particular diagnostics generated by the invention can trigger specific actions. Several examples follow: phones calls can be made, machines can be turned on to run specific tasks, computer programs can be triggered and run.
In the same manner, particular diagnostics generated by the invention can trigger a routine in the invention, generating new tests for data analysis, modifying formulas or erasing previous tests in the invention. In this manner, the invention becomes a learning mechanism, under the teachings of artificial intelligence.
Therefore, the invention represents cost reductions in the development of many types of large and complicated data-analysis applications. The resulting applications are easily customizable by the user.
Further objects and advantages of the invention will become apparent in the consideration of the drawings and ensuing description.
SUMMARY OF THE INVENTION
Definition:
Expert systems are intelligent computer programs that use knowledge and inference procedures to solve problems that are hard enough as to require, in their solution, significant expertise.
Expert system building tools typically comprise (a) special data structures for the storage of knowledge, (b) a so-called inference engine for the logical operation of knowledge through the inference process, (c) interface mechanisms for the user to define goals and data, when required, (d) mechanisms for handling uncertainty and missing data and knowledge. All these mechanisms are designed in a form most efficient and appropriate to the operation of expert systems.
The invention combines expert system building tools with database management instruments (a) applying expert knowledge to any information contained in databases and reports, (b) linking the expert system results, hereby called diagnostics, with selected related data on the databases and reports, as defined by the user, and (c) presenting a query interface capable of structuring and showing the resulting diagnostics and the original linked data according to user preferences such as actions and priorities.
Structure:
The invention's components are, as shown in FIG. 2, and also in FIG. 1 and FIGS. 3 to 8: data database (5), tests database (9), test-processing engine (10), diagnostics database (7), querying engine (8) and inference engine (6).
The data database (5), as shown in FIG. 3, may contain one or more of the following: (A) data tables (13) and related index files (16) for identifying each individual data item (12), (B) group elements data tables (15) that state which data items correspond to each group, so as to perform a common test on all these data items, and (C) group descriptions data tables (14).
The tests database (9), as shown in FIG. 4, may contain one or more of the following: (A) tests descriptions (17) and associated formulas data tables (20), (B) data items identifiers (21) or group identifiers identifying the set of items to which the test will be applied (22), and (C) triggers data table (23), where each test is associated to a particular set of triggers, and one or more triggers may produce an action (24), where an action is any activity defined by the user.
The tests-processing engine (10), as shown in FIG. 5, is capable of interpreting each if-then-else test (18 and 20), group (22) and data item identifiers (21) and triggers (23), so as to produce the diagnostics (7) corresponding to the data database (5).
The diagnostics database (7), as shown in FIG. 6, may contain one or more diagnostics data tables (27). Each record in these data tables will contain two sets of fields. One set of fields contains identifiers (18) of those tests that produced the TRUE condition, and the other set of fields contains data item identifiers (21) for the particular data items where the tests produced the TRUE condition. Triggers (23) and their associated actions (24) may be reached immediately in the test database (9), through the associated index files (16).
The querying engine (8), as shown in FIG. 8, is capable of sorting (33), filtering (34), linking (35) and showing synchronized database browse views (36) of the previous three databases, according to the user selections and preferences.
The expert system and its associated inference engine (6), as shown in FIG. 7, comprise a knowledge database (28), sets of rules (29), a rules processing language (30), inference logic procedures (31), and interface modules (32).
The invention may be implemented in different manners. In certain implementations, tests processing language would be the same. However, some practical implementations will keep these separate in order to present a simpler operating interface to the user. The description of the preferred embodiment will show these principles more clearly.
Operation:
In general, to operate, this invention proceeds through three logical stages: (1) It creates a relational database with information retrieved from different computer sources. (2) Through use of test processing procedures and, optionally, an expert system building tool, applications built with it emulate expert reasoning processes to associate data with concepts, and it calculates and evaluates indexes to generate diagnostics databases. (3) It permits users to query these diagnostics databases.
This invention can be used to develop a wide variety of applications in areas as diverse as financial analysis, organizational evaluation, and project control. As it has already been explained, the preferred embodiment of the invention will be mostly applied for the development of executive information systems. In the following paragraphs, we will primarily refer to the particular example where this invention embodies an executive information system (EIS).
At the first logical step outlined previously, in the case where this invention embodies an EIS, the relational database retrieves information from the firm's distributed information system, which could be a network of personal computers, workstations, mainframes and minicomputers. This invention, like other typical current database query systems, will be able to access information from a wide variety of computers and data formats, and thus it should be able to build the first relational database from a set of queries from all these sources. The nature of these queries shouldn't be a constraint to this invention, given that the originality of the invention lies in the novel and useful combination of the database, the expert system, the building of the diagnostics database and the coordinated query system thereof. However, in the best embodiment, the first database could be built in a manner similar to a database reporting system. That is, just as a typical reporting system generates a set of printed reports from the queries over the distributed database, this invention's queries generate a set of tables as files in the computer disk or memory. Remember that all reports are really printed tables, that is, tables that are later formatted and sent to a different device.
However, this first step may be implemented in an entirely different way. For example, a driver or set of drivers may allow the program the direct access of all distributed databases in the firm, as tables. In that case, the first step could almost be eliminated, since no new data files need be created: the data database (5) wouldn't be necessary, and the business database (1 to 4) would take its place. Such a scheme would be more efficient in terms of database storage space, but less efficient in terms of program complexity and speed of operation of the expert system--the next logical step--.
At the second logical step outlined previously, an expert system emulating human reasoning processes to associate data with concepts calculates and evaluates indexes to generate a diagnostics database. If, for example, the databases contain financial information and financial ratios, the expert system will generate the diagnostics associated to those financial ratios. If the company's debt ratio is too high, the expert system could generate a diagnostic database entry stating that "the company's debt is to large". This diagnostic statement could later be analyzed in subsequent passed of the rule system, together with all other diagnostic statements generated from the firm's financial statements, so as to suggest strategies and priorities for management.
This invention will also work if instead of the of an expert system, (a) a simpler language with a simpler set of if--Then logical condition statements and (b) an engine consisting of an appropriate set of test-processing procedures are provided. This particular case will be presented in the best embodiment implementation. Such a system could also be completely integrated with other expert systems and inference engines.
Every diagnostic generated by the expert system or the if--Then condition statements engine may be linked to the data it originated from. The links between particular data items in the database and the corresponding diagnostic items in the diagnostics database may be bi-directional or unidirectional, depending on the particular application and user preferences. In a bi-directional linking scheme, it's possible to go from every particular data item to the diagnostics that are linked to it, and it's also possible to go each diagnostic to the corresponding data item(s) this diagnostic has originated from.
These linkings may be defined by the user, or they may also be created automatically. If the links are created automatically, they simply connect every diagnostic with each one of all the data items needed to calculate the logical/mathematical test in the diagnostic. On the other hand, many times it's preferable to let the user define what data items to link each diagnostic to. For example, if the quantitative database contains numbers for sales, costs of sales and profits, and a particular diagnostic test checks if profits are going down, then quite probably the user would prefer to link this resulting diagnostic to the sales numbers, to the costs numbers and to the profits numbers, so that he'll be reminded that the profits are going down when looking at the sales or at the costs figures. Another example: if the quantitative database contains numbers for total debt (liabilities), total assets and the debt ratio (which calculates as debt/assets), then if the debt ratio is too high (for example, higher than 0.50) a diagnostic should be issued indicating "careful: the debt is too high". Again, the user will probably want to link this diagnostic to the assets numbers, to the debt numbers and to the profit ratio numbers, so that he'll be reminded that debt is too high when looking at the assets or at the liabilities figures.
Finally, at the third logical step outlined previously, this invention lets users query, in a synchronized manner, the diagnostics database and the original data database.
The invention's design lets the user do this easily: the user only needs to point the cursor at any data item, and he'll simultaneously see the key diagnostics associated with it in a separate window in the screen.
For example, FIG. 9 shows a snapshot from a computer screen. The upper half shows a few concepts from a fictional Balance Sheet for ABC Inc. The lower half of the screen shows the diagnostics associated to the data item selected by the cursor. FIG. 20 shows a similar screen, as it has been developed in the best mode implementation.
The invention stores the diagnostics in a parallel database which allows the user to sort and organize all the diagnostics information in any way he likes.
For example, with a single keystroke, the user can call a "speed bar" with the names of the diagnostics as shown in FIG. 10. Notice that only the highlighted cells meet the diagnostic's criteria. The user may sort the information in any way that he may choose to, such as by department, by product or service, or by year. The user can scroll through the diagnostics, or through the application's original data. Whenever the user scrolls through the diagnostics, he will see, in a synchronized manner, the associated data from which these diagnostics originated, and vice versa.
The invention incorporates downdrilling techniques to navigate through the data. While looking at any screen, the user only needs to point at the desired cell, press ENTER, and a more detailed report relating to the corresponding data will instantly appear on the monitor. The invention may use these downdrilling techniques on information based on the diagnostics, on the original data, or on actions (also defined by the user) that should ensue whenever a predefined test and its associated diagnostic are met.
FIGURES (brief description of drawings)
FIG. 1 represents a general but simplified view of the invention's components: the organization's--i.e., firm or institution--distributed computers and databases (1 to 4), the invention's data database (5), the expert system with its inference engine (6), the diagnostics database (7) and the querying engine (8). The invention proceeds through three logical stages: First, it creates a data database (5) with information retrieved from different computer sources (1 to 4). Second, through use of an expert system building tool (6), it generates a diagnostics database (7). Third, the querying engine (8) allows users to query these diagnostics databases.
FIG. 2 shows the general components in the best embodiment implementation: The invention's data database (5) which is built from the information on the organization's databases, the tests database (9), the test-processing engine (10), the expert system with its inference engine (6), the diagnostics database (7) and the querying engine (8). Notice the differences when compared with the structure in FIG. 1: The organization's databases (1 to 4) aren't shown, since these are non-essential to the invention's operation and the invention's data table contains all their information. Two components are added: the tests database (9) and the test-processing engine (10). Inference engines and expert systems (6) typically contain a tests database and test-processing engine, but in the best-embodiment implementation, an additional set of these components is provided.
FIG. 3 shows the data database's (5) basic components: one or more database tables containing lines or records (11) containing individual data items (12) of the data to be analyzed (13), group descriptions data tables (14), group elements data tables (15) listing the individual data items (12), and/or lines or records (11) of the data that comprise each group, and related index files (15).
FIG. 4 shows the tests database's (9) components: the tests descriptions table (17) containing the test identifier (18), name and description for each test (19), the associated formula for each test (20), identifiers for individual data items in the data database (21) on which the tests will be applied; group identifiers (22) of the data items in the data database on which the tests will be applied; the triggers data table (23) and their associated actions database (24) and related index files (25).
FIG. 5 shows a simplified outline of the tests-processing engine's (10) operation: Each test's formula (20) is read from the test database (9). The test-processing procedures (26) apply this formula to the corresponding data database's (5) data items, according to the data database's data item identifiers (21) and the group identifiers (22) associated to each test. When the result of any of these test's formulas (20) is TRUE, some information associated to that particular test is written to the diagnostics database (7). The TRUE test's--i.e., the diagnostic's--information includes: the test identifier (18), the data item identifiers (21) associated to the particular formula turned TRUE, and the associated triggers (23) and actions (24).
FIG. 6 shows the diagnostics database's (7) contents: diagnostics data tables (27) containing, for each test that turned TRUE--i.e. for each diagnostic--, the test identifier (18), the data item identifiers (21), the associated triggers (23) and actions (24).
FIG. 7 shows the expert system and associated inference engine (6). This instrument contains a knowledge database (28), one or more sets of rules (29), a rules processing language (30), inference logic procedures (31) so as to develop a heuristic sequence of steps to achieve the stated goals, and several interface modules (32) for the user to input his data, new rules, or to respond to the expert system's questions. These elements don't need to be described more thoroughly here, since expert systems have been, for the past ten years, a widely known type of artificial intelligence application system in the computer science community.
FIG. 8 shows the querying engine (8), a set of procedures for sorting (33), filtering (34), linking index tables (35) associated to data database (5), the tests database (9), and the diagnostics database (7), and showing synchronized database browse views (36) of the data database (5), tests database (9), and diagnostics database (7).
FIG. 9 shows a typical query browse view of the data database (5) and the associated individual diagnostics from the diagnostics database (7). Before looking at this screen, the user executed the test-processing engine (10) to apply all tests in the tests database (9) to all data items in the data database (5). The screen's cursor is positioned on `545.1`, a data database's data item (12) representing the current assets for February 1992. The lower part of the screen shows five diagnostics associated to that particular data item.
FIG. 10 shows the data database (5), one associated diagnostic from the diagnostics database (7) at the bottom of the screen, and a list of test identifiers (18) at the right edge of the screen. The user first selected `538.8`, a data database's data item (12), then the user pressed a key to switch cursor action to the list of test identifiers. In FIG. 10, the active cursor is positioned on CURROK, a test identifier. Those data items where the CURROK test turned TRUE (517.4, 558.9, 86.9, 86.4) are highlighted and underlined. If the user moves the cursor from CURROK to any other test identifier, the highlighted data items will change. Highlighted data items always correspond to the data items where the test identifier selected by the cursor turned TRUE.
FIG. 11 shows a summary diagnostics screen. The invention querying engine (8) is capable of generating this type of reports. The titles at the top and left side of the table (such as `high priority`, `banks`) are the triggers (23) and actions (24) associated to all diagnostics in the diagnostics database (7). The numbers in the table represent the number or frequency counts of diagnostics associated to each combination of triggers and actions. The user may point with a pointing device such as a mouse or keyboard cursor, and thus select any number in the table, or any title at the top or left side of the table. In FIG. 11, the cursor is positioned on `high priority`.
FIG. 12 shows a summary diagnostics screen. This is the screen the invention could show right after executing the cursor selection (`high priority`) in FIG. 11. This shows more detail on the high priority actions and triggers.
FIG. 13 shows the same summary diagnostics screen in FIG. 11, but the cursor selection is now positioned on number `12`, which, when executed by pressing a key, would select the `low priority` and `CALL` actions and triggers.
FIG. 14 shows a summary diagnostics screen. This is the screen the invention could show right after executing the cursor selection (`low priority` and `CALL`) in FIG. 13. This shows more detail on these actions and triggers.
FIG. 15 shows a sample data database (5). This example contains the numbers of a small balance sheet summary and a few associated financial ratios.
FIGS. 16, 17 and 18 show partial listings of the preferred embodiment's menu interface. File menu options let the user select an application and which files he will operate on, importing data from other applications and exporting it back to them, and printing options. Edit menu options let the user browse, look at, modify and add information in any of the individual files of the database in the invention's preferred embodiment implementation. The diagnostics menu options let the user create new tests and formulas, apply tests to selected data items on the data database (5), and execute the querying engine (8) as to query the diagnostics database (7) and the data database (5) in a coordinated manner. Notice there are four browse options in the diagnostics menu: data browse view, diagnostics browse view, frequency browse view, summary browse view. These are described in more detail in FIGS. 20 to 36.
FIG. 19 shows the preferred embodiment's applications picklist. The invention is capable of handling many types of applications. The preferred embodiment allows the user to select an application to execute.
FIG. 20 shows a sample screen of the best embodiment's data browse view. This browse view shows the data table in the upper half of the screen. By moving with the cursor, the user can move over the data table's contents. The lower half of the screen shows a listing of those diagnostics associated to the data item selected by the cursor at any time. This is the most convenient browse option to use when the user prefers to look at his original firm's data, and looking at the associated diagnostics at the same time. In FIG. 20, the cursor is positioned on `101.53`, a data item (12) representing the value of `Current Liabilities ($)` for September, 1993. As shown, the user may select, with the cursor, any individual data item in the data database (5). The cursor's current position is signaled on the screen by the underline and highlight. Two diagnostics that correspond to that particular data item are shown on the lower part of the screen.
FIG. 21 shows a sample screen of the best embodiment's data browse view. All browse screens generated by the browse options in the diagnostics menu are composed of two tables, at the upper and lower half of the screen and, optionally, may also show at any time a more detailed description corresponding to the diagnostic selected by the cursor. In all browse view screens, cursor control can either be in the table at the upper half of the screen, or in the table at the lower half of the screen. The user can switch cursor control over any of the two tables by pressing a special key. In the preferred embodiment's implementation, this special key is the F5 key. Thus, the F5 key acts like a switch, transferring control between the upper and lower portions of the screen. The screen shown in FIG. 21 is reached after pressing the F5 key in the screen shown in FIG. 20. The cursor is now positioned on the first diagnostic, identified by the ".cndot." symbol, and by underlined and highlighted text. The user can scroll over the diagnostics in this list. At the right of this list, the screen shows a longer description of the diagnostic selected by the cursor at any time. In FIG. 21, this longer description reads: "RDEB.sub.+ --The debt ratio is too--low (lower than 0.25). You may be losing business opportunities, or your company's accounting practices mix several businesses in one accounting practice".
FIG. 22 shows a sample screen of the best embodiment's diagnostics browse view. This browse view shows the diagnostics in the upper half of the screen. By moving with the cursor, the user can move over the diagnostics. The lower half of the screen shows the data item associated to the diagnostic selected by the cursor at any time. This is the most convenient browse option to use when the user prefers to look at the diagnostics, but when it's also important to know where these diagnostics have come from. Also, this browse view is appropriate when there are lots of data and not too many diagnostics to look at. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 23 shows a sample screen of the best embodiment's diagnostics browse view. The screen shown in FIG. 23 is reached after pressing the F5 key in the screen shown in FIG. 22. The cursor is now positioned on the line identified by the ".cndot." symbol, and by underlined and highlighted text. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 24 shows a sample screen of the best embodiment's frequency browse view. This browse view shows, in the upper half of the screen, the main diagnostics classes. These correspond to the summary description's triggers and associated actions data tables (23, 24). If the user previously selected to look at a particular case in a class, the invention will show the classification in the following class. The lower half of the screen shows the diagnostics corresponding to the class or group selected by the cursor at any time. This is the most convenient browse option to use when the user prefers to define priorities, actions, individual persons' responsibilities, or any other classification of diagnostics according to their classes as defined by the user. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 25 shows a sample screen of the best embodiment's frequency browse view. The screen shown in FIG. 25 is reached after pressing the F5 key in the screen shown in FIG. 24. The cursor is now positioned on the line identified by the ".cndot." symbol, and by underlined and highlighted text. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 26 shows a sample screen of the best embodiment's summary browse view. By definition, in the invention a summary diagnostic turns ON only when a predefined set of individual diagnostics have turned ON before. For example, a summary diagnostic may verify that sales and profits have gone down and costs have gone up for five different products, and would tell the manager that there's a crisis with the financial margins' of the firm. This browse view shows, in the upper half of the screen, those summary diagnostics that turned TRUE. By moving with the cursor, the user can move over the summary diagnostics. The lower half of the screen shows the individual diagnostics corresponding to the summary diagnostic selected by the cursor at any time. This is the most convenient browse option to use for looking at the overall picture according to the diagnostics. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 27 shows a sample screen of the best embodiment's summary browse view. The screen shown in FIG. 27 is reached after pressing the F5 key in the screen shown in FIG. 26. The cursor is now positioned on the line identified by the ".cndot." symbol, and by underlined and highlighted text. Refer to FIGS. 20 and 21 for a more complete description of the diagnostic browse options and screens.
FIG. 28 shows the data selection screen for the best embodiment's frequency browse view. In fact, the four browse view options in the diagnostics menu use the same data selection screen. In order not to be forced to navigate through all diagnostics or all through data in each browse view option, the user may select which lines, classes or periods he will limit the search to. The program filters data using an AND combination of all selected fields. There's only one exception: the line option is not a filter, but a go-to switch that makes the browse view go to the selected line but will always show all data database's lines.
FIG. 29 shows the data selection screen for the frequency browse view as described in FIG. 28. The user is selecting an option for class1, one of the trigger fields (23).
FIG. 30 shows the frequency browse view the user would get after making the choice shown in FIG. 29. The list of diagnostics and the related frequencies refer only to the selection on class1: `weakness--low priority`. Notice that the program's frequency list shows the classification by the triggers in class2.
FIG. 31 shows the data selection screen for the `frequency browse view`, as described in FIG. 28. The user is selecting an option for class2, one of the trigger fields.
FIG. 32 shows the frequency browse view the user would get after making the choice shown in FIG. 31. The list of diagnostics and the related frequencies refer only to the selection on class2: `Sales--D. Vadim (Dan)`. Notice that the program's frequency list shows the classification by the triggers in class1.
FIG. 33 shows the data selection screen for the frequency browse view, as described in FIG. 28. The user is selecting an option for class3, one of the trigger fields.
FIG. 34 shows the frequency browse view the user would get after making the choice shown in FIG. 33. The list of diagnostics and the related frequencies refer only to the selection on class3: `BANKS`. Notice that the program's frequency list shows the classification by the triggers in class1.
FIG. 35 shows the data selection screen for the frequency browse view, as described in FIG. 28. The user has selected an option for class1 (`ALARM1`) and for class3 (`BANKS`), two of the trigger fields.
FIG. 36 shows the frequency browse view the user would get after making the choice shown in FIG. 35. The list of diagnostics and the related frequencies refer to the selections on class1 AND (logical AND) class3 (`weakness--low priority`, and `prepare appropriate banking arrangements`). Notice that the program's frequency list shows the classification by the triggers in class2, the only class for which a selection hasn't been made yet.
FIG. 37 shows a listing of typical line tests and formulas for the best embodiment implementation. This listing only shows partial contents of the test database. The full contents (fields, type, size) of each record in this database file is described in FIG. 46. FIGS. 47 and 48 show two snapshots of the full contents of a few selected records. Let's come back to FIG. 37. The formulas in FIG. 37 mostly compare the financial ratios in the data database with absolute references that may be established for the firm. For example, the first line after the headings in each table means that, if the Z1 score falls below 0.15 (as shown in the formula), that indicates extreme danger of bankruptcy (as told in the test.sub.-- description). In the best embodiment implementation shown here, the data database contains 60 months of periodic information (September '93, October '93, etc.). For practical purposes, the formulas may be applied to any of these periods in exactly the same manner. The identification of each data item is of the form F("1.BALANCE.18",0), where " 1.BALANCE.18" is the line or record identifier, and 0 is a relative period identifier: it means current data period (-1 means previous data period, etc.). The user tells the program which period or periods he wishes the tests performed on. Then, the program will define each one of those periods as the current period. It will do it one period at a time. Each one of these formulas can only be applied to one set of data items, identified in the formula. And, if the formula turns out to be TRUE, then a diagnostic is generated and written to the diagnostics database. That diagnostic only needs to identify the test that originated it and the data item to which it refers.
FIG. 38 shows a listing of typical group tests and formulas. Refer to FIG. 37 for a related description on line formulas. The group tests shown here have been selected mostly for their clarity, not because they are the most useful and practical tests in this case. In this example, there are four groups, which are ASSETS, LIABIL, EQUITY, and FRATIO. One group formula is applied to each group. Applying a group formula to a group means applying that group formula to all data items in that group. The typical data identifier in each group formula looks like G(0), where `0` means CURRENT data period. No identifier for a line or record is needed, since all group formulas can only be applied to one line at a time, and can only compare different elements within the same line. Thus, group formulas are applied one line at a time, for all lines that belong to the particular group referred to by the group formula. In this example, group formulas can only be used for inter temporal comparisons. It is fairly easy to extend this standard as to allow for more complete variations to these formulas.
FIG. 39 shows the preferred embodiment's implementation menu structure: The File menu contains options to select an application and which files he will operate on, importing data from other applications and exporting it back to them, and printing options. The Edit menu contains options to browse, look at, modify and add information in any individual database. The Diagnostics menu contains options to create new tests and formulas, apply tests to selected data items, and query the diagnostics database and the data database in a coordinated manner. The Maintenance menu contains options for the generation of index files, file backups, and other maintenance operations on the data. The Options menu contains password and security options, options for the selection of the initial period, other options controlling the display and printout formats. The Help menu contains options for activating and modifying context help screens.
FIG. 40 shows the general structure of the best embodiment's file structure. There are twelve essential DBF files for each application. Conceptually, these files may be organized in three groups: data files (table.dbf, which contains the data to be analyzed; grupos.dbf and detalle.dbf, which contains the data items grouping definitions), test files and diagnostic files (for simple diagnostics and for summary diagnostics). For more detailed information on the whole file structure, see FIGS. 73 and 74.
FIG. 41 shows the structure of the data database file and its contents (13) in the best embodiment's implementation. This database file is called TABLA.DBF, and the fields contained in each record are called: LINEA (49), NOMBRE (50), P1 to P60 (51), VALOR (68), FORMAT (69), FACTOR (70), TYFACE (71). LINEA (49) contains the data line or record identifier (11 ), NOMBRE (50) contains the data item description, P1 to P60 (51) contain the actual data items (12), in this case, simple numbers, VALOR (68) contains a temporary value only used for importing and exporting data for other applications--this is non-essential to the invention and to this patent application--, FORMAT (69), FACTOR (70) and TYFACE (71) contain formatting information for each data item--non-essential here--.
FIGS. 42 and 43 show partial browse views of database file TABLA.DBF, as described in FIG. 41.
FIG. 44 shows the structure of the diagnostics database file and its contents (13) in the best embodiment's implementation. This database file is called RESULT.DBF, and the fields contained in each record are called: LINEA (49), NPER (52), PRUEBA (53), CLASE1 (61), CLASE2 (61), CLASE3 (61). LINEA (49) contains the data line or record identifier (11), NPER contains the period identifier for the related data item (12), PRUEBA (53) contains the test's identifier (18), CLASE1 (61), CLASE2 and CLASE3 contain the triggers' identifiers (23). Important note: the diagnostic identifiers are the same as the test identifiers (18), since a test turned true is automatically called a diagnostic.
FIG. 45 shows a partial browse view of database file RESULT.DBF, as described in FIG. 44.
FIG. 46 shows the structure of the tests database file (9) in the best embodiment's implementation. This database file is called PRUEBA.DBF, and the fields contained in each record are called: CODIGO (54), NOMBRE (55), LINEA (49), GRUPO (58), CONDEXPR (57), CONDNOTE (56), CLASE1 (61), CLASE2 (61), CLASE3 (61). CODIGO (54) contains the test identifier (18), NOMBRE (55)contains the test's short description (17), LINEA (49)contains the identifier (21) for the data database's line or record to which the resulting diagnostic will be related in case the test turns true, GRUPO (58) contains the identifier for the associated group of data items on which the test should be performed, CONDEXPR (57) contains the test's formula (20), CONDNOTE contains a long description for the test (17), CLASE1 (61), CLASE2 and CLASE3 contain the associated triggers and actions (23, 24). Important note: the diagnostic identifiers are the same as the test identifiers (18), since a test turned true is automatically called a diagnostic.
FIGS. 47, 48 and 49 show partial browse views of database file PRUEBA.DBF, as described in FIG. 46.
FIG. 50 shows the structure of the groups' definitions database file (14) in the best embodiment's implementation. This database file is called GRUPOS.DBF, and the fields contained in each record are called: CODIGO (59), NOMBRE (60). CODIGO (59) contains the groups' identifier (22), NOMBRE (60) contains the groups' descriptions (14).
FIG. 51 shows a partial browse view of database file GRUPOS.DBF, as described in FIG. 50.
FIG. 52 shows the structure of the groups' elements database file (15) in the best embodiment's implementation. This database file is called DETALLE.DBF, and the fields contained in each record are called: GRUPO (58), LINEA (49). GRUPO (58) contains the groups' identifiers (22), LINEA (49) contains the identifier for the data database's line or record (21) that added to each particular group.
FIG. 53 shows a partial browse view of database file DETALLE.DBF, as described in FIG. 52.
FIG. 54 shows the structure of the triggers database file (23, 24) in the best embodiment's implementation. This database file is called CLASE1.DBF, and the fields contained in each record are called: CLASE1 (61), NOMBRE1 (62). CLASE1 (61) contains the triggers identifiers (23, 24), NOMBRE1 (62) contains the triggers descriptions. The preferred embodiment's implementation also contains two files called CLASE2.DBF and CLASE3.DBF. Their structure is exactly the same as in file CLASE1.DBF. Their contents are completely definable according to the user's needs. No additional explanation is needed.
FIG. 55 shows a partial browse view of database file CLASE1.DBF, as described in FIG. 54.
FIG. 56 shows the structure of the summary test definition database file in the best embodiment's implementation. This database file is called PRESUMEN.DBF, and the fields contained in each record are called: ANALISIS (63), NOMBRE (64). CONDNOTE (65), PMIN (66). ANALISIS (63) contains the summary test identifiers, NOMBRE (64) contains the summary tests descriptions, CONDNOTE (65) contains summary tests' short descriptions, and PMIN (66) contains the minimum number of individual diagnostics that will turn a summary test on.
FIGS. 57 and 58 show partial browse views of database file PRESUMEN.DBF, as described in FIG. 56.
FIG. 59 shows the structure of the summary tests elements database file in the best embodiment's implementation. This database file is called DRESUMEN.DBF, and the fields contained in each record are called ANALISIS (63), PRUEBA (53). ANALISIS (63) contains the summary test identifiers, PRUEBA (53) contains the diagnostics identifiers.
FIG. 60 shows a partial browse view of database file DRESUMEN.DBF, as described in FIG. 58.
FIG. 61 shows the structure of the summary diagnostics database file i.e., those summary tests that have turned out to be true, in the best embodiment's implementation. This database file is called PRRESULT.DBF, and the fields contained in each record are called: ANALISIS (63), NPER (52). ANALISIS (63) contains the summary test's identifier, NPER (52) contains the period number identifier of the summary diagnostic (i.e., for those periods where the summary test turned out to be true).
FIG. 62 show partial browse views of database file PRRESULT.DBF, as described in FIG. 61.
FIG. 63 shows the structure of the summary diagnostic elements database file in the best embodiment's implementation. This file lists the individual diagnostics for those summary tests that turned out to be true. This database file is called DRRESULT.DBF, and the fields contained in each record are called: ANALISIS (63), NPER (52), PRUEBA (53), LINEA (49), POSITIVA (67). ANALISIS (63) contains the summary test's identifier, NPER (52) contains the period number identifier, PRUEBA (53) contains the individual tests identifiers (i.e., the identifiers for the components or individual elements in the summary test), LINEA (49) contains the identifier for the associated data database's lines or records, POSITIVA (67) contains a single character indicator which can either be "P" when an individual test turned true or blank when not.
FIG. 64 show padial browse views of database file DRRESULT.DBF, as described in FIG. 63.
FIG. 65 shows the structure of the files database file in the best embodiment's implementation. This database file is non-essential to the invention, and is only necessary for consistency checking in this particular implementation. This file is called IBASES.DBF, and the fields contained in each record are called: NOMBRE, BYTES, RECORD, MENSAJE, CARACTER.
FIG. 66 show partial browse views of database file IBASES.DBF, as described in FIG. 65.
FIG. 67 shows the overall flow diagram of the best embodiment's implementation: FIRST, the invention must access all the firm's databases (see 1 to 4 in FIG. 1) to build the data database (5). SECOND, the invention generates a complete database structure and related index files (16). The just-generated test and diagnostic database files (7) are empty, but ready to receive information. The detailed database structure, index files contents and relations between the files will be made evident in the next Figures of this application. THIRD, define data groups. This is potentially one of the most powerful features of the invention. The user can define any groups (14, 15) of data items (12), and even groups of groups, in a multilevel structure. These groups are useful for the application of common tests to all data items in any particular group, for the definition of reports and queries, and for the definition of the associated triggers and actions (23, 24). FOURTH, define tests (17, 18, 19) and formulas (20, 21, 22). Formulas can be as simple as IF--THEN tests applied to one or more data items or data groups, or as sophisticated as expert system clauses and rules can be. FIFTH, apply formulas. The test-processing engine (10) then interprets and executes all formulas, reading and comparing all relevant data items, and whenever a test's conditions turn TRUE, generating one or more diagnostic records in the diagnostics database (7). Any diagnostic reference stored in the diagnostic database will also contain an associated reference to one or more data items (12). These data items will be defined by the user, or they will simply be the data items examined when the test condition turned true and the ensuing diagnostic was generated. SIXTH, query the application. The user may query the diagnostics database and the data database simultaneously in two or more separate windows, linked through several index files (16).
FIG. 68 shows the steps required to use and query an application: (a) Select the desired application. Use command File--Application. (b) Observe selected tables (optional). Use command Edit--Browse records. Observe selected tables from the application's database (5+7+9). This is optional. The user will do it only if he/she wants to. (c) Browse the summary diagnostics. Use command Diagnostics--Summary browse view. Observe or browse the summary diagnostics, for example, to understand the general condition of the firm. (d) Browse the diagnostic frequencies. Use command Diagnostics--Frequency browse view. Observe or browse the diagnostic frequencies classified by appropriate triggers and actions, according to the user preferences. (e) Browse the individual diagnostics. Use command Diagnostics--Diagnostics browse view. Observe or browse the diagnostics trying different trigger and action combinations (class1, class2, class3) to identify the original data related to key diagnostics. (f) Browse the data and diagnostics. Use command Diagnostics--Data Browse view. Observe or browse the original data for a final review, since this is the option closest to traditional report-analysis methods. (g) Select another application (optional). Use command File--Application.
FIG. 69 shows the steps required to create an application: (a) Select the desired application. Use command File--Application. (b) Create first line tests. Use command Diagnostics--Add/modify Line test. Create the first line tests for a simple application. (c) Run tests. Use command Diagnostics--Run tests. Run all tests for the set of periods desired.
FIG. 70 shows the steps required to create an application with group tests: (a) Follow steps shown in FIG. 69, steps to create an application. (b) Select the groups file. Use command File--Open database (grupos.dbf). Select the groups database file. (c) Enter groups codes and descriptions. Edit--Add records (grupos.dbf). Enter the codes and descriptions for all lines and data items groups the user wishes to create to later operate with group formulas. (d) Select groups detail file. Use command Use command File--Open database (detafie.dbf). Select the groups database detail file (detalle.dbf). (e) Define groups elements. Use command Edit--Add records (detafie.dbf). Enter the identifiers for all data database's lines and data items belonging to each group recently defined. (f) Create group tests. Use command Diagnostics--Add/modify Group test. Create the group tests desired (review the samples presented in this document).
FIG. 71 shows the steps required to create an application with group tests and triggers: (a) Follow steps shown in FIG. 69: steps to create an application. (b) Now, follow steps shown in FIG. 70. optional steps to create an application with group tests. (c) Select the triggers file. Use command File--Open database (clasel.dbf). Select the triggers and actions (class1) database file. (d) Define triggers information. Use command Edit--Add records (clasel.dbf). Define all codes and descriptions for the triggers and actions (class1) that you will use in the application (for example, strengths and weaknesses, high and low priorities, departments, responsible persons). (e) Define other triggers (optional). Define other triggers and actions (optional). Replicate the procedure followed for file class1, with class2 (file clase2. dbf) and class3 (file clase3.dbf). (f) Select the tests file. Use command File--Open database (pruebas.dbf). Select the tests database file (pruebas.dbf). (g) Enter triggers identifiers for each test. Use command Edit--Add records (pruebas.dbf). Enter all codes (identifiers) for those triggers and actions that were defined (clase1, clase2, clase3), and for each test created.
FIG. 72 shows the steps required to create an application with summary tests and diagnostics: (a) Follow steps shown in FIG. 69: steps to create an application. (b) Now, follow steps shown in FIG. 70: optional steps to create an application with group tests. (c) Then, follow steps shown in FIG. 71: optional steps to create an application with group tests and triggers. (d) Select summary tests definition file. Use command File--Open database (resumen.dbf). Select the summary tests definition database file (resumen.dbf). (e) Define summary tests information. Use command Edit--Add records (resurnen.dbf). Define codes and descriptions for all summary tests, and the minimum number of diagnostics required for each summary diagnostic to show. (f) Select summary test elements file. Use command File--Open database (drresumen.dbf). Select the summary tests detail database file (drresumen.dbf). (g) Define summary tests elements. Use command Edit--Add records (drresumen.dbf). Define all tests identifiers (i.e., diagnostics) that should be examined in each summary test.
FIG. 73 shows a diagram of the best implementation's database (DBF) files structure. It shows the files' select numbers as they are consistently used in the best embodiment's code, and it graphically shows the relations (one to one, one to many, many to many) in the relational file structure. (Also see FIG. 74.)
The data DBF file (37) is identified in the program's code by Select 1: tabla.dbf. This file contains the data to be analyzed (13), as described in the summary of the invention.
The diagnostics DBF file (38) is identified in the program's code by Select 2: result.dbf. This file corresponds, in the best embodiment, to the diagnostics data tables (27) described in the summary of the invention.
The tests DBF file (39) is identified in the program's code by Select 3: prueba.dbf. This file corresponds to the tests database (9) and its component elements (17 to 24) as described in the summary of the invention.
The group descriptions DBF file (40) is identified in the program's code by Select 4: grupos.dbf. This file corresponds to the group descriptions data tables (14) described in the summary of the invention.
The group elements DBF file (41) is identified in the program's code by Select 5: detalle.dbf. This file corresponds to the group elements data tables (15) described in the summary of the invention.
The clase1 DBF file (42) is identified in the program's code by Select 6: clase1.dbf. This file corresponds to the triggers and associated actions data tables (23, 24) described in the summary of the invention.
The clase2 DBF file (43) is identified in the program's code by Select 7: clase2.dbf. This file corresponds to the triggers and associated actions data tables (23, 24) described in the summary of the invention.
The clase3 DBF file (44) is identified in the program's code by Select 8: clase3.dbf. This file corresponds to the triggers and associated actions data tables (23, 24) described in the summary of the invention.
The summary test DBF file (45) is identified in the program's code by Select 9: presumen.dbf. This part of the tests database (9) wasn't separately in the summary description of the invention.
The summary detail DBF file (46) is identified in the program's code by Select 10: dresumen.dbf. This is part of the tests database (9), although it wasn't explicitly described in the summary of the invention.
The summary diagnostics DBF file (47) is identified in the program's code by Select 11: prresult.dbf. This file corresponds to the diagnostics data tables (27) described in the summary of the invention.
The summary diagnostics elements DBF file (48) is identified in the program's code by Select 12: drresult.dbf. This file corresponds to the diagnostics data tables (27) described in the summary of the invention.
FIG. 74 shows the FoxPro.TM. database language by Microsoft's set view listing of best embodiment's database relations. This option can be reached by selecting windows-view from FoxPro.TM. 2.5's main menu. FIG. 74 shows the default database files relations structure. This is changed frequently during execution, but reset again to this structure.
FIG. 75 shows a schematic view of the first Data browse view screen. Compare it with FIG. 20. The cursor is active in the data browse window (72) showing selected contents from the data database (5), i.e. the best mode's data DBF file (37) tabla.dbf. The diagnostics browse window (73) shows selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf.
FIG. 76 shows a schematic view of the second Data browse view screen. Compare it with FIG. 21. The cursor is active in the diagnostics browse window (75), showing selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The data browse window (74) shows related records from the data database (5), i.e. the best mode's data DBF file (37) tabla.dbf. The text window (76) shows the test description (19) related to the active record in the diagnostics database (7).
FIG. 77 shows a schematic view of the third Data browse view screen. Compare it with FIG. 21. The cursor is active in the diagnostics browse window (78), showing selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The data browse window (77) shows related records from the data database (5), i.e. the best mode's data DBF file (37) tabla.dbf. The text window (79) shows the test description (19) related to the active record in the diagnostics database (7).
FIG. 78 shows a schematic view of the first Diagnostics browse view screen. Compare it with FIG. 22. The cursor is active in the diagnostics browse window (80), showing selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The data browse window (81) shows related records from the data database (5), i.e. the best mode's data DBF file (37) tabla.dbf.
FIG. 79 shows a schematic view of the second Diagnostics browse view screen. Compare it with FIG. 23. The cursor is active in the data browse window (83), showing selected contents from the data database (5), i.e. the best mode's data DBF file (37) tabla.dbf. The diagnostics browse window (82) shows related records from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf.
FIG. 80 shows a schematic view of the first Frequency browse view screen. Compare it with FIG. 24. The cursor is active in the frequencies browse window (84). The diagnostics browse window (85) shows related records from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf.
FIG. 81 shows a schematic view of the second Frequency browse view screen. Compare it with FIG. 25. The cursor is active in the diagnostics browse window (87), showing selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The frequencies browse window (86) shows related records from a temporary diagnostics file (for immediate frequency counts). The text window (88) shows the test description (19) related to the active record in the diagnostics database (7).
FIG. 82 shows a schematic view of the first Summary browse view screen. Compare it with FIG. 26. The cursor is active in the summaries browse window (89), showing selected contents from the summary diagnostics DBF file (47) prresult.dbf. The diagnostics browse window (90) shows related records from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The text window (91) shows the summary test description related to the active record in the summary test DBF file (45) presumen.dbf.
FIG. 83 shows a schematic view of the second Summary browse view screen. Compare it with FIG. 27. The cursor is active in the diagnostics browse window (93), showing selected contents from the diagnostics database (7), i.e. the best mode's diagnostics DBF file (38) result.dbf. The summaries browse window (92) shows related records the summary diagnostics DBF file (47) prresult.dbf. The text window (94) shows the test description (19) related to the active record in the diagnostics database (7).
FIG. 84 shows procedure Apply.sub.-- formulas as it is executed in the best embodiment implementation. This procedure executes four routines: first, the structure for all files involved must be defined (95), then this procedure executes line tests and formulas (96), group tests and formulas (97) and summary tests (98). These three routines are always executed one immediately after the other, and in that order. The program modules involved are: vapl.prg for line tests, vaplg.prg for group tests and vaplr.prg for summary tests. For more details about these modules and their associated algorithms, see Section C below: description of formula application procedures.
FIG. 85 shows procedure Data.sub.-- browse.sub.-- view. This represents the algorithm to generate the screens shown in FIGS. 20 and 21. This algorithm is described in more detail below in Section D: Description of diagnostic browse operations, and the associated schematic Figures are 75 to 77. First, procedure (99) initializes all variables and procedure (100) selects the data DBF file tabla.dbf (37), the diagnostics DBF file result.dbf (38) and the test DBF file prueba.dbf (39). These files can now be accessed by commands Select 1, 2 and 3, respectively. Now procedure (101) defines windows tabla, result and result1. These windows are described in FIGS. 75 to 77. Then, procedure (102) allows the user to choose which filters to apply on the data, using the filter screen described in FIGS. 31 to 36. Procedure (103) shows the first coordinated browses view as shown in FIG. 20 and described in FIG. 75. If the user presses the F5 key, procedure (104) will execute and generate the second coordinated browses view as shown in FIG. 21 and described in FIG. 76. If the user presses the F6 key, procedure (105) will show the third coordinated browses view, which is described in FIG. 77.
FIG. 86 shows procedure Diagnostics.sub.-- browse.sub.-- view. This represents the algorithm to generate the screens shown in FIGS. 22 and 23. This algorithm is described in more detail below in Section D: Description of diagnostic browse operations, and the associated schematic Figures are 78 and 79. First, procedure (106) initializes all variables and procedure (107) selects the diagnostics DBF file result.dbf (38), the data DBF file tablaodbf (37) and the test DBF file prueba.dbf (39). These files can now be accessed by commands Select 2, 1 and 3, respectively. Now procedure (108) defines windows tabla and result. These windows are described in FIGS. 78 and 79. Then, procedure (109) allows the user to choose which filters to apply on the data, using the filter screen described in FIGS. 31 to 36. Procedure (110) shows the first coordinated browses view as shown in FIG. 22 and described in FIG. 78. If the user presses the F5 key, procedure (101) will execute and generate the show second coordinated browses view as shown in FIG. 23 and described in FIG. 79.
FIG. 87 shows procedure Frequencies.sub.-- browse.sub.-- view. This represents the algorithm to generate the screens shown in FIGS. 24 and 25. This algorithm is described in more detail below in Section. D: Description of diagnostic browse operations, and the associated schematic Figures are 80 and 81. First, procedure (112) initializes all variables and procedure (113) selects the diagnostics DBF file result.dbf (38), the test DBF file prueba.dbf (39), the temporary DBF file temp1.dbf and the temporary DBF file temp2.dbf. File temp.dbf will contain the results of the frequency count operations. File temp1.dbf will contain a selection of the diagnostics DBF file result.dbf (38). Now procedure (114) defines windows tabla, result and result1. These windows are described in FIGS. 80 and 81. Then, procedure (115) allows the user to choose which filters to apply on the data, using the filter screen described in FIGS. 31 to 36. Procedure (116) shows the first coordinated browses view as shown in FIG. 24 and described in FIG. 80. If the user presses the F5 key, procedure (117) will execute and generate the show second coordinated browses view as shown in FIG. 25 and described in FIG. 81.
FIG. 88 shows procedure Summary.sub.-- browse.sub.-- view. This represents the algorithm to generate the screens shown in FIGS. 26 and 27. This algorithm is described in more detail below in Section D: Description of diagnostic browse operations, and the associated schematic FIGS. are 82 and 83. First, procedure (118) initializes all variables and procedure (119) selects the summary tests DBF file presumen.dbf (45), the summary tests elements DBF file dresumen.dbf (46) and the summary diagnostics DBF file prresult.dbf (47). These files can now be accessed by commands Select 9, 10 and 11, respectively. Now procedure (120) defines windows tabla, result and result1. These windows are described in FIGS. 82 and 83. Then, procedure (121) allows the user to choose which filters to apply on the data, using the filter screen described in FIGS. 31 to 36. Procedure (122) shows the first coordinated browses view as shown in FIG. 26 and described in FIG. 82. If the user presses the F5 key, procedure (123) will execute and generate the show second coordinated browses view as shown in FIG. 27 and described in FIG. 83.
FIG. 89 shows an outline of the most important code routines used in the best-mode implementation of the invention. There are menu administration routines to create and activate the main menus, to call the main menu routines and to call the diagnostic menu routines. The corresponding program source files are called vacambio.prg, vmenu.prg and vacti.prg. Similarly, vdrive.prg and vverif.prg let the user select an application by chaing the application path default, and check the structure of each file. Vcond.prg, vcondg.prg and vdup.prg let the user add, modify or duplicate line and group tests. Vapl.prg, vaplg.prg and vaplr.prg apply the logic tests to the quantitative database, adding new records to the diagnostics database whenever the corresponding tests turn out to be true. Vman.prg, vmanr.prg, vresu.prg and vsumm.prg are the four main browse views that let the user look at the diagnostics and the quantitative data in a coordinated manner, according to the links defined previously. Finally, vbases.prg, vlista.prg and vano.prg provide maintenance services to the program.
FIG. 90 shows the basic operation of the invention, which involves the selection of an item or items of quantitative data and the application of the corresponding analysis rule(s) to generate diagnostic records. The linking pointers are added to the new diagnostic records and this information is stored in the diagnostic database. In the best embodiment implementation, all these operations are performed in the following manner, as exemplified in FIG. 90: First, the program makes the first record in the test database the current test record (124). It must be noted that each test record contains a test code unique to this particular test record, and a description of the mathematical calculation(s) and logic operation(s) to be performed. The program now retrieves the necesary data needed for the test encoded in the test formula in the current test record (125). Then the mathematical calculation(s) and logic operation(s) in the first test are performed (126). After these operations, if the result of the test just applied is true, a new record in the diagnostic database is created (128), and the following information is written into this new diagnostic record: the keyword identifying the particular data item and a test code identifying the test record and encoding the result of the test. Finally, if this is not the last test record in the test database, that is, if there are more tests to perform, as tested by block 129, then the next test record is selected from the test database (130), and the procedure is repeated again starting at step (125), retrieving the necessary data needed for the corresponding test just selected.
FIG. 91 shows, in a simplified manner, the procedure followed to display and select a particular data item and to see or browse through the diagnostics related to the particular data item selected. The sequence of steps is: (a) display window 1 and window 2 in the computer screen; (b) display one or more items of data from quantitative database in window 1; (c) display at least one diagnostic record from the diagnostics database in window 2; (d) display a pointer in window 1; (e) receive user commands to move the pointer and enter a select command so as to select an item of data displayed in window 1; (f) retrieve from the diagnostic database the diagnostic records associated with the selected item of data selected in window 1 using the link pointer programmed by the user when the test which generated the diagnostic was programmed by the user, and (g) display the selected diagnostic records in window 2.
FIG. 92 shows, in a simplified manner, the procedure followed to display and select a particular data item and to see/browse through the diagnostics related to the particular data item selected. The sequence of steps is: (a) display window 1 and window 2 in the computer screen; (b) display one or more items of data from quantitative database in window 1; (c) display at least one diagnostic record from the diagnostics database in window 2; (d) display a pointer in window 2; (e) receive commands from the user to move the pointer and issue a select command to select a diagnostic record displayed in window 2; (f) retrieve from the quantitative database the data records associated with the diagnostic record selected in window 2 using the link pointer programmed by the user when the test which generated the diagnostic was programmed by the user, and (g) display the selected quantitative data records in window 1.
FIG. 93 shows a four-way browse view where four simultaneous browse views are shown. Just like in the best embodiment, here each browse view is the result of a particular BROWSE command from the FoxPro.TM. language. The four browse views mentioned are identified in the figure by the titles: "1 expert window", "2 actions window", "3 diagnostics window", "4 data window". There are also two more windows, called "keywords window" and "text window". Here, if the user selects the "4 data window" by use of some pointing mechanism, the program will then operate in a manner similar to that shown in FIG. 20. That is, the user may scroll through the data and the associated diagnostics will appear on window "3. diagnostics window". Similarly, if the user selects the "3 diagnostics window", the program will operate in a manner similar to that described in FIG. 22. If the user selects the "2 actions window", the program will operate in a manner similar to that described in FIG. 24, the frequencies view. If the user selects the "3 expert window", the program will operate in a manner similar to that described in FIG. 26, the summaries view. Also, the keywords window is really a filter window, and it operates in a manner similar to the screen shown in FIG. 28. Finally, the text window is the same window shown at the lower right corner of FIG. 21, and showing the full text of the description of a particular selected diagnostic.
PREFERRED EMBODIMENT
Structure:
A list of all parts identified by reference numerals in FIGS. 1 to 89 follows:
1. to
4. business database
5. data database
6. inference engine and expert system
7. diagnostics database
8. querying engine
9. tests database
10. test-processing engine
11. lines or records containing individual data items
12. individual data items of the data to be analyzed
13. data to be analyzed
14. group descriptions data tables
15. group elements data tables
16. related index files
17. tests descriptions table
18. test identifiers
19. name and description for each test
20. associated formulas
21 data database's data item identifiers
22. group identifiers
23. triggers data table
24. associated actions database
25. related index files
26. test-processing procedures
27. diagnostics data tables
28. knowledge database
29. sets of rules
30. rules processing language
31. inference logic procedures
32. interface modules
33. sorting procedures
34. filtering procedures
35. linking procedures
36. procedures for showing synchronized database browse views
37. data DBF file (select 1: tabla.dbf)--this file contains the data to be analyzed (13)
38. diagnostics DBF file (select 2: result.dbf)--this file corresponds to the diagnostics data tables (27)
39. tests DBF file (select 3: prueba.dbf)--this file corresponds to the tests database (9) and its components
40. group descriptions DBF file (select 4: grupos.dbt)--this file corresponds to the group descriptions data tables (14)
41. group elements DBF file (select 5: detalle.dbf)--this file corresponds to the group elements data tables (15)
42. clase1 DBF file (select 6: clase1.dbf)--this file corresponds to the triggers and actions data tables (23, 24)
43. clase2 DBF file (select 7: clase2.dbf)--this file corresponds to the triggers and actions data tables (23, 24)
44. clase3 DBF file (select 8: clase3. dbf)--this file corresponds to the triggers and actions data tables (23, 24)
45. summary test DBF file (select 9: presumen.dbf)--this file is a part of the tests database (9)
46. summary elements DBF file (select 10: dresumen.dbf)
47. summary diagnostics DBF file (select 11: prresult.dbf)--this file corresponds to the diagnostics data tables (27)
48. summary diagnostics elements DBF file (select 12: drresult.dbf)--this file corresponds to the diagnostics data tables (27)
49. data record identifier--field: LINEA--files: tabla.dbf, result.dbf, prueba.dbf, detalle.dbf, drresult.dbf
50. data record description--field: NOMBRE--files: tabla.dbf
51. data fields--fields: P1 to P60--files: tabla.dbf
52. data field identifier--field: NPER--files: result.dbf, prresult.dbf, drresult.dbf
53. test identifier--field: PRUEBA--files: result.dbf, dresumen.dbf, drresult.dbf
54. test identifier--field: CODIGO--files: prueba.dbf. Same as test identifier (53)
55. test description--field: NOMBRE--files: prueba.dbf
56. long test description--field: CONDNOTE--files: prueba.dbf
57. test formula--field: CONDEXPR--files: prueba.dbf
58. group identifier--field: GRUPO--files: prueba.dbf, detalle.dbf
59. group identifier--field: CODIGO--files: grupos.dbf. Same as group identifier (58)
60. group description--field: NOMBRE--files: grupos.dbf
61. class identifier--fields: CLASE1, CLASE2, CLASE3--files: result.dbf, prueba.dbf, clase1.dbf, clase2.dbf, clase3.dbf
62. class description--fields: NOMBRE1, NOMBRE2, NOMBRE3--files: clase1.dbf, clase2.dbf, clase3.dbf
63. summary test identifier--field: ANALISIS--files: presumen.dbf, dresumen.dbf, prresult.dbf, drresult.dbf
64, summary test description--field: NOMBRE--files: presumen.dbf
65. summary test long description--field: CONDNOTE--files: presumen.dbf
66. minimum number of positive tests--PMIN--files: presumen.dbf
67. test occurrence flag--field: POSITIVA--files: drresult.dbf
68. non-essential field: temporary data--field: VALOR--files: tabla.dbf
69. non-essential field: formatting information 1--field: FORMAT--files: tabla.dbf
70. non-essential field: formatting information 2--field: FACTOR--files: tabla.dbf
71. non-essential field: formatting information 3--field: TYFACE.--files: tabla.dbf
72. data browse window of the first Data browse view screen
73. diagnostics browse window of the first Data browse view screen
74. data browse window of the second Data browse view screen
75. diagnostics browse window of the second Data browse view screen
76. text window of the second Data browse view screen
77. data browse window of the third Data browse view screen
78. diagnostics browse window of the third Data browse view screen
79. text window of the third Data browse view screen
80. diagnostics browse window of the first Diagnostics browse view screen
81. data browse window of the first Diagnostics browse view screen
82. diagnostics browse window of the second Diagnostics browse view screen
83. data browse window of the second Diagnostics browse view screen
84. frequencies browse window of the first Frequency browse view screen
85. diagnostics browse window of the first Frequency browse view screen
86. frequencies browse window of the second Frequency browse view screen
87. diagnostics browse window of the second Frequency browse view screen
88. text window of the second Frequency browse view screen
89. summaries browse window of the first Summary browse view screen
90. diagnostics browse window of the first Summary browse view screen
91. text window of the first Summary browse view screen
92. summaries browse window of the second Summary browse view screen
93. diagnostics browse window of the second Summary browse view screen
94. text window of the second Summary browse view screen
95. define file relations (procedure Apply.sub.-- formulas)
96. apply line formulas (procedure Apply.sub.-- formulas)
97. apply group formulas (procedure Apply.sub.-- formulas)
98. apply summary formulas (procedure Apply.sub.-- formulas)
99. setup screen options (procedure Data.sub.-- browse.sub.-- view)
100. activate files (procedure Data.sub.-- browse.sub.-- view)
101. define windows in screen (procedure Data.sub.-- browse.sub.-- view)
102. filters selection (procedure Data.sub.-- browse.sub.-- view)
103. show first coordinated browses view (procedure Data.sub.-- browse.sub.-- view)
104. show second coordinated browses view (procedure Data.sub.-- browse.sub.-- view)
105. show third coordinated browses view (procedure Data.sub.-- browse.sub.-- view)
106. setup screen options (procedure Diagnostics.sub.-- browse.sub.-- view)
107. activate files (procedure Diagnostics.sub.-- browse.sub.-- view)
108. define windows in screen (procedure Diagnostics.sub.-- browse.sub.-- view)
109. filters selection (procedure Diagnostics.sub.-- browse.sub.-- view)
110. show first coordinated browses view (procedure Diagnostics.sub.-- browse.sub.-- view)
111. show second coordinated browses view (procedure Diagnostics.sub.-- browse.sub.-- view)
112. setup screen options (procedure Frequencies.sub.-- browse.sub.-- view)
113. activate files (procedure Frequencies.sub.-- browse.sub.-- view)
114. define windows in screen (procedure Frequencies.sub.-- browse.sub.-- view)
115. filters selection (procedure Frequencies.sub.-- browse.sub.-- view)
116. show first coordinated browses view (procedure Frequencies.sub.-- browse.sub.-- view)
117. show second coordinated browses view (procedure Frequencies.sub.-- browse.sub.-- view)
118. setup screen options (procedure Summary.sub.-- browse.sub.-- view)
119. activate files (procedure Summary.sub.-- browse.sub.-- view)
120. define windows in screen (procedure Summary.sub.-- browse.sub.-- view)
121. filters selection (procedure Summary.sub.-- browse.sub.-- view)
122. show first coordinated browses view (procedure Summary.sub.-- browse.sub.-- view)
123. show second coordinated browses view (procedure Summary.sub.-- browse.sub.-- view)
124. make the first record in the test database the current test record
125. retrieve the necesary data needed for the test encoded in the test formula in the current test record
126. perform the mathematical calculation(s) and logic operation(s) in the first test
127. check if the result of the test just applied is true
128. create a new record in the diagnostic database and write to it the keyword identifying the particular data item and a test code identifying the test record
129. check if this is the last test record in the test database
130. select the next test
Referring to FIG. 2, the best mode implementation's entire program storage area is divided in the data database (5), the tests database (9), the test-processing engine (10), the inference engine and expert system (6), the diagnostics database (7) and the querying engine (8). Inference engines and expert systems (6) typically contain a tests database and test-processing engine, but in the best-embodiment implementation, an additional set of these components is provided.
There are twelve essential DBF files for each application, as shown in FIG. 40. Conceptually, these files may be organized in three groups: data files (table.dbf, which contains the data to be analyzed; grupos.dbf and detalle.dbf, which contains the data items grouping definitions), test files and diagnostic files (for simple diagnostics and for summary diagnostics).
FIG. 74 shows Foxpro.TM.'s set view listing of these files' relations. FIG. 73 shows these relations in more detail. It also shows the files' select numbers as they are consistently used in the best embodiment's code.
The data DBF file (37) is identified in the program's code by Select 1: tabla.dbf. This file contains the data to be analyzed (13), as described in the summary of the invention. Also see FIGS. 42 and 43.
The diagnostics DBF file (38) is identified in the program's code by Select 2: result.dbf. This file corresponds, in the best embodiment, to the diagnostics data tables (27) described in the summary of the invention. Also see FIGS. 44 and 45.
The tests DBF file (39) is identified in the program's code by Select 3: prueba.dbf. This file corresponds to the tests database (9) and its component elements (17 to 24) as described in the summary of the invention. Also see FIGS. 46 to 49.
The group descriptions DBF file (40) is identified in the program's code by Select 4: grupos.dbf. This file corresponds to the group descriptions data tables (14) described in the summary of the invention. Also see FIGS. 50 and 51.
The group elements DBF file (41) is identified in the program's code by Select 5: detalle.dbf. This file corresponds to the group elements data tables (15) described in the summary of the invention. Also see FIGS. 52 and 53.
The clase1, clase2 and clase3 DBF files (42, 43, 44) are identified in the program's code by Select 6, 7 and 8: clase#.dbf. This files correspond to the triggers and associated actions data tables (23, 24) described in the summary of the invention. Also see FIGS. 54 and 55.
The summary test DBF file (45) is identified in the program's code by Select 9: presumen.dbf. This part of the tests database (9) wasn't separately in the summary description of the invention. Also see FIGS. 56 to 58.
The summary detail DBF file (46) is identified in the program's code by Select 10: dresumen.dbf. This is part of the tests database (9), although it wasn't explicitly described in the summary of the invention. Also see FIGS. 59 and 60.
The summary diagnostics DBF file (47) is identified in the program's code by Select 11: prresult.dbf. This file corresponds to the diagnostics data tables (27) described in the summary of the invention. Also see FIGS. 60 and 61.
The summary diagnostics elements DBF file (48) is identified in the program's code by Select 12: drresult.dbf. This file corresponds to the diagnostics data tables (27) described in the summary of the invention. Also see FIGS. 61 and 62.
The invention's best embodiment implementation has been designed to run as a stand-alone system, or in an integrated manner with an expert system and its associated inference engine. As stated previously, researchers define expert systems in the following way: Intelligent computer programs that use knowledge and inference procedures to solve problems that are hard enough as to require in their solution, significant expertise. The expert system and its associated inference engine (6), as shown in FIG. 7, comprise a knowledge database (28), sets of rules (29), a rules processing language (30), inference logic procedures (31), and interface modules (32).
Operation:
As shown in FIG. 67, after building the data table and the associated empty files with the structure described in the previous paragraphs, in broad terms, the user will (a) define the tests to perform on the data, either in the test database or in the associated expert system, (b) will apply these tests and generate the associated diagnostics and classes, i.e. triggers and actions, and (c) the user will finally be able to query the diagnostics and the data in a coordinated manner.
FIG. 15 shows an excerpt of a typical application data table. FIG. 41 shows the data table's fields structure, and FIG. 42 shows another partial browse view of this table. In the best embodiment's data table, also called data DBF file (37) each record represents a different concept, such as sales, costs or profits, and each field represents a different time period. In a typical case, each field will represent a different consecutive month.
The user will usually start his work with the data by defining data item groups. These groups are useful for the application of common tests to all data items in any particular group, for the definition of reports and queries, and for the definition of the associated triggers and actions. FIG. 52 shows the structure of the group elements DBF file detalle.dbf (41), and FIG. 53 shows a partial browse view of this sample file. In this file, groups are defined by a simple list of pairs: the group identifier CODIGO and the data record identifier LINEA. Remember that, to completely identify a data item, only the record and field are needed, but as it may be seen in the data file's structure, data fields correspond to data periods. The user will later select the CURRENT data period on which all operations will be made.
The group definition process is complete when filling out the information required in the group descriptions DBF file grupos.dbf (40). FIG. 50 shows the fields structure of this file, and FIG. 51 shows a partial sample browse view of the same file.
Now the user will define tests and formulas. It will help now to know that, later, when the program applies user's tests on the data, it will do so only for a particular data period at a time. Thus, the user may tell the invention to work, for example, only on the data on September '93, which could be data field P7. Then, that data period, or field P7, would become period 0. Period 0 is the formula reference for CURRENT period. The previous period is then referred as -1, the period previous to that as -2, etc.
Formulas can be as simple as IF-THEN tests applied to data items, but they may become as sophisticated as expert system rules can be. The tests DBF file prueba.dbf holds both line tests and group tests. Line tests contain a record reference in field LINEA, while group tests contain a group reference in field GRUPO. FIG. 46 shows the file's fields structure. FIGS. 47 and 48 show two snapshots of the full contents of a few selected records.
FIG. 37 shows a listing of typical line tests and formulas of the IF-THEN type. The identification of each data item is of the form F("1.BALANCE.18",0), where "1.BALANCE.18" is the line or record identifier, and 0 is a relative period identifier: it means current data period (-1 means previous data period, etc.). The identification of each data item is of the form F("1.BALANCE.18",0), where "1.BALANCE.18" is the line or record identifier, and 0 is a relative period identifier: it means current data period (-1 means previous data period, etc.). If the formula turns out to be TRUE, then a diagnostic is generated. That diagnostic only needs to identify the test that originated it (test.sub.-- identifier, field PRUEBA) and the data item to which it refers, referred by the record identifier, field LINEA.
FIG. 38 shows a listing of typical group tests and formulas. Applying a group formula to a group means applying that group formula to all data items in that group. The typical data identifier in each group formula looks like G(0), where 0 means current data period. No identifier for a line or record is needed, since all group formulas can only be applied to one line at a time, and can only compare different elements within the same line. Thus group formulas may be used for inter temporal comparisons.
The test-processing engine (10) then interprets and executes all formulas for a particular data period at a time, reading and comparing all relevant data items, and whenever a test's conditions turn TRUE, generating one or more diagnostic records in the diagnostics DBF file result.dbf (38).
FIG. 44 shows the diagnostics DBF file's structure, and FIG. 45 shows a sample partial browse view of the same file. Whenever a test turns out to be true, a new record is added on the diagnostics DBF file. Each of these records simply stores two fields: the test identifier, and the data item identifier.
The test identifier is field PRUEBA in file result.dbf, which is read from field CODIGO in file prueba.dbf. In xBase notation (or FoxPro.TM. notation), this is written as RESULT.PRUEBA=PRUEBA.CODIGO. The data item identifier is composed of two fields: the data record identifier, which is field LINEA in file data.dbf, and the CURRENT period identifier, which is stored in field NPER of the file diagnostic DBF file result.dbf. Remember that the CURRENT period identifier had to be determined by the user just before running all tests.
The procedures executed to apply line and group tests to the data are described in all detail below in Section C: Description of formula application procedures.
The test DBF file prueba.dbf also holds fields CLASE1, CLASE2 and CLASE3. The operation and meaning of these fields can best be understood by reviewing FIGS. 29 to 36, and their associated descriptions. These fields serve several purposes: they are a means to classify all tests, and hence their derivative diagnostics; these classifications may represent anything the user fancies, such as the people responsible, whether the tests and diagnostics represent opportunities, dangers, etc., what should be done about each one of them, etc. These fields can also be interpreted as triggers and actions (23, 42, 43, 44), since whenever a test containing a particular code in a CLASE# field turns true, a particular action, procedure or program may be executed. The code to develop such an application should be obvious to anyone skilled in the art.
After the diagnostics have been generated, the user may query the diagnostics database and the data database simultaneously in two or more separate windows. The mapping of the data into the diagnostics combined with the coordinated query of both the data and the derivative diagnostics are the qualities that make this invention particularly innovative. The fact that the formulas may be an expert system's rules, and the diagnostics its own associated expert diagnoses only point out to the flexibility of this arrangement. The fact that no other implementation generating and querying diagnostics in any way similar to this, at least in functionality, has ever appeared in the market, only underlines the uniqueness and unobviousness of the invention.
The query procedures produce at least four different types of coordinated browse views, which may be seen in FIGS. 20 to 27. It is also recommended to study the associated descriptions.
The algorithms that produce each one of these four coordinated browse views are explained in detail below in Section D: description of diagnostic browse operations. The operation of the frequency browse view is more complicated since two temporary DBF files, temp.dbf and temp1.dbf have to be generated to hold the frequency counts for each class of diagnostics, depending on the filters applied on the query, as shown in FIGS. 29 to 36, and to hold the individual diagnostics associated to each frequency count and to each class of diagnostics defined in the query.
The operation of the summary browse view also involves a new set of files to hold and define the summary tests and diagnostics.
The summary diagnostics are capable of representing the overall condition of the problem under analysis by the invention. They can even foretell dangers, opportunities and scenarios for the near term future.
For example, a firm may have the following goals: (a) yearly 20% sustained growth on sales, (b) cost growth at no more than 10% per year, and (c) reducing the debt ratio to no more than 45%. In this case, the user may ask The invention to show him when these three conditions are met in five products or more.
Summary diagnostics are turned on only when the individual diagnostics in their definition are ON simultaneously.
Individual diagnostics that have turned ON are identified by letter P. Those individual diagnostics not identified with letter P have not turned on. If those individual diagnostics not identified with letter P happened to turn TRUE, then all diagnostics in the summary diagnostic would be TRUE.
Summary diagnostics could, because of their nature, also be named checklist diagnostics.
Summary diagnostics are defined in the summary elements DBF file dresumen.dbf (46). The field structure for this file is shown in FIG. 60. FIG. 61 shows a partial sample browse view of the same file. To define a summary test, the user only needs to define a set of data pairs: the summary test identifier, and an identifier for another individual diagnostic that must be true for the summary test to be true too. The summary test identifier is field ANALISIS (63), and the individual diagnostic is field PRUEBA (53). To complete the summary test definition, the user may also write a longer description for the test in file presumen.dbf (45), with the structure shown in FIGS. 58 and 59. The user may also define the minimum number of individual diagnostics that will set the summary diagnostic on.
Finally, the summary diagnostics are stored in files prresult.dbf and drresult.dbf in a manner similar to the storage of individual diagnostics and summary tests. For more details, see the description for the associated procedures in Section D below.
It's also necessary to describe the linking procedure. It can be clearly understood from the rest of the best-embodiment's description, particularly that of files tabla.dbf (containing the quantitative data), prueba.dbf (containing the tests and its associated formulas) and result.dbf (containing the diagnostics, which are those tests that have turned out to be true for specific data items).
Each particular data item needs may be identified by identifying the data record and the data field it's read from. Field LINEA is used to identify each data record in the quantitative database. The field identifier (1 to 60, since there are sixty period numbers) is used to identify the particular data field where the data item is located. In this discussion, we will call this field identifier "period number".
Each test is identified by its code CODIGO. Finally, individual diagnostics ARE NOT identified by any particular code. The way the linking is done is simple: whenever a particular test identified by CODIGO turns out to be true, then a new record is added to the diagnostic database (PRUEBA.DBF). The new diagnostic record will contain the test identifier CODIGO, which will now be stored in field PRUEBA, and it will also contain the identifier for one particular data item the diagnostic will be linked to (fields LINEA and NPER for period number identifier).
In this way, each diagnostic record will contain the test code (PRUEBA) and the data item identifier (LINEA, NPER). Thus, when looking at a particular diagnostic, it's easy to find the related (linked) test and the related (linked) data item. On the other hand, when looking at a particular data item, it's relatively easy to find all those diagnostics that are related to it by simply searching for the data item identifier (with the corresponding values for LINEA and NPER) in all diagnostic records. This search operation could take some time. In order to speed up the search operation, the files may be indexed following database indexing procedures commonly used in the industry: a single statement INDEX should be issued, indicating which fields (LINEA, CODIGO, PRUEBA, NPER) the index is going to be built on. The rest is automatically done by the database administrator. In the case of the invention's best embodiment, the FoxPro.TM. 2.0 database manager was used.
FIG. 90 shows the basic operation of the invention, which involves the selection of an item or items of quantitative data and the application of the corresponding analysis rule(s) to generate diagnostic records. The linking pointers are added to the new diagnostic records and this information is stored in the diagnostic database. In the best embodiment implementation, all these operations are performed in the following manner, as exemplified in FIG. 90: First, the program makes the first record in the test database the current test record (124). It must be noted that each test record contains a test code unique to this particular test record, and a description of the mathematical calculation(s) and logic operation(s) to be performed. The program now retrieves the necesary data needed for the test encoded in the test formula in the current test record (125). Then the mathematical calculation(s) and logic operation(s) in the first test are performed (126). After these operations, if the result of the test just applied is true, a new record in the diagnostic database is created (128), and the following information is written into this new diagnostic record: the keyword identifying the particular data item and a test code identifying the test record and encoding the result of the test. Finally, if this is not the last test record in the test database, that is, if there are more tests to perform, as tested by block 129, then the next test record is selected from the test database (130), and the procedure is repeated again starting at step (125), retrieving the necessary data needed for the corresponding test just selected.
FIG. 91 shows, in a simplified manner, the procedure followed to display and select a particular data item and to see or browse through the diagnostics related to the particular data item selected. The sequence of steps is: (a) display window 1 and window 2 in the computer screen; (b) display one or more items of data from quantitative database in window 1; (c) display at least one diagnostic record from the diagnostics database in window 2; (d) display a pointer in window 1; (e) receive user commands to move the pointer and enter a select command so as to select an item of data displayed in window 1; (f) retrieve from the diagnostic database the diagnostic records associated with the selected item of data selected in window 1 using the link pointer programmed by the user when the test which generated the diagnostic was programmed by the user, and (g) display the selected diagnostic records in window 2.
FIG. 92 shows, in a simplified manner, the procedure followed to display and select a particular data item and to see/browse through the diagnostics related to the particular data item selected. The sequence of steps is: (a) display window 1 and window 2 in the computer screen; (b) display one or more items of data from quantitative database in window 1; (c) display at least one diagnostic record from the diagnostics database in window 2; (d) display a pointer in window 2; (e) receive commands from the user to move the pointer and issue a select command to select a diagnostic record displayed in window 2; (f) retrieve from the quantitative database the data records associated with the diagnostic record selected in window 2 using the link pointer programmed by the user when the test which generated the diagnostic was programmed by the user, and (g) display the selected quantitative data records in window 1.
Ramifications and Scope of Invention
While the previous description contains many specificities, these should not be construed as limitations on the scope of the invention, but rather as an exemplification of one preferred implementation thereof. The following paragraphs explain many other alternative schemes:
If the previous description of the preferred embodiment is compared to the description of Expert Systems in section. (B) under Background of the Invention, it can be seen that in one possible implementation, this invention may be implemented as an add-in tool to existing database programming languages or to Expert System programming languages and development environments. In that case, this invention would add specific facilities for creating a database from the diagnostics generated by the expert system's programming statements. Other additions or modifications that could be done to the invention include: (A) drilldown reporting, (B) the transparent, automatic and synchronized operation of tests and diagnostics to data in any kind of computer application, even through concurrent programming operations or multitasking environments, (C) different data structures for the data DBF file, allowing the easy operation of case-oriented applications such as high-speed real-time industrial processes, (D) recursive expert analysis techniques where the tests themselves become data for the expert system and the test-processing engine, and (E) different schemes for integrating the invention with expert systems, neural networks. Most of these implementations will be briefly discussed here.
A. Referring to DRILLDOWN REPORTING:
Drilldown reporting is a feature particularly important to managers and all those people that need to work with summary information but still reach the details that originate from those summaries and at the same time explain them. With drilldown reporting, a manager must be able to look at a very abbreviated summary report, showing only grand totals of all the operation. By pointing at any particular data in the summary, the drilldown program will immediately show the detail components of the particular data element pointed to. Many different levels may be defined, and the drilldown-drillup process goes only one step at a time. Thus, from grand totals, the user may go to totals, partial detail, and full detail. Then, the user may go back up again to the totals.
The best mode implementation defines drilldown reporting by defining groups of lines, and defining classes within tests and diagnostics. Whenever the user chooses any particular combination of line filters and classes filters, he automatically gets a summary report in which he may downdrill as much as he wants by looking at the summary data in one screen and the detail screen in the other (like in the summary browse view, and the frequency browse view, which show totals and summaries in the upper half of the screen, and the detail information in the lower half). A more sophisticated implementation of this scheme would allow the recursive definition of groups. Currently, groups can only be composed of individual data items, i.e. records or lines in the tabla.dbf file. Also, classes can only be composed of individual class items. In an expanded version, groups may also contain groups as long as a strict hierarchy of groups is respected, that is, no circularities are involved, since these would most probably result in logical contradictions. Groups containing groups and classes containing classes would mean an unlimited possible number of downdrilling levels. The user would define a summary report by simply listing the groups, classes and periods (or groups of periods) that should show on the vertical and horizontal axis of the summary report, and the filter that should be applied to all the data on that report. From then on, whenever the user selects any data item on the report, the next subsequent group in the hierarchical definition scheme would be displayed in the report instead. For an example of this class of application, see again FIGS. 11 to 14 and their associated descriptions.
Also, as shown on FIG. 93, the particular way the browse screens and browse view are created and shown in the preferred embodiment isn't the best way to do this type of analysis. FIG. 93 shows a four-way browse view where four simultaneous browse views are shown. Just like in the best embodiment, here each browse view is the result of a particular BROWSE command from the FoxPro.TM. language. The four browse views mentioned are identified in the figure by the titles: "1 expert window", "2 actions window", "3 diagnostics window", "4 data window". There are also two more windows, called "keywords window" and "text window". Here, if the user selects the "4 data window" by use of some pointing mechanism, the program will then operate in a manner similar to that shown in FIG. 20. That is, the user may scroll through the data and the associated diagnostics will appear on window "3. diagnostics window". Similarly, if the user selects the "3 diagnostics window", the program will operate in a manner similar to that described in FIG. 22. If the user selects the "2 actions window", the program will operate in a manner similar to that described in FIG. 24, the frequencies view. If the user selects the "3 expert window", the program will operate in a manner similar to that described in FIG. 26, the summaries view. Also, the keywords window is really a filter window, and it operates in a manner similar to the screen shown in FIG. 28. Finally, the text window is the same window shown at the lower right corner of FIG. 21, and showing the full text of the description of a particular selected diagnostic. The advantage of the arrangement shown in FIG. 93 over those described in the best embodiment is that here all windows will be active at all times. That is, the user only needs to point at a particular window to find a particular action (or class), a particular expert diagnostic (or summary diagnostic, as it was called during the description of the best embodiment), etc. The user will be able to directly jump from the expert (summary) diagnostics to the associated quantitative data, etc.
B. Referring to the INTEGRATION with other software APPLICATIONS: the invention can also be applied to current implementations of accounting programs, spreadsheet applications, statistical analysis packages, etc. This could be implemented through several ways. In the simplest possible way, the invention remains as described in the best embodiment. That is, all tests are evaluated on the data contained in the tabla.dbf file. However, the tabla.dbf file could contain extracts of particular data from these programs, obtained earlier by reading it from these programs. Then, whenever the invention executes the coordinated diagnostic browse view operations, the invention would work as described in the best mode implementation but, instead of showing the data DBF file tabla.dbf, the invention would show the original data from the associated program. In order to do this, the associated program would only need to dynamically generate and send to the invention two indexes to identify the field and record position of the corresponding data item in the data DBF file tabla.dbf. That is, an index file linking the application's screens and tabla.dbf would have to be generated. Multitasking environments such as the Microsoft Windows.TM. environment by Microsoft Corporation of Redmond, Wash. state, allows the simultaneous operation of several applications in separate open and visible windows, and the dynamic sending of live data between these applications, through what are called DDE (dynamic-data-exchange) channels. Schemes such as this could also allow the simultaneous execution of all types of applications and the invention. This would be important to managers and users of all kinds, since it would allow them to look at the data in a different way, organized by their own priorities and responsibilities, and the ensuing actions.
C. Referring to DIFFERENT DATA STRUCTURES for the data DBF file: the best embodiment implementation as shown here is appropriate for time-based applications such as the analysis of financial reports (balance sheets, financial ratios, profit and loss statements, etc.). For case-oriented applications such as high-speed real-time industrial processes, quality control applications, inventory applications for the control of individual transactions and items, etc., a different data DBF file structure should be proposed. In the best embodiment, records represent concepts such as sales and costs, and fields represent time periods. In an alternative case-oriented scheme, fields would be individual cases, which could well be repetitive instances of the same readings. Each field would represent a particular element of a reading. For example, in a toy manufacturing plant, the statistics for each particular toy manufactured may be stored in a database. The reading for serial product #10358 could be: time 3:58 minutes, plastics grade A 12.15 grams, weight 8.15. There could be many other data elements. Every three seconds a new reading for a new manufactured toy could be generated, adding a new record with those three fields (and probably many more) to the data DBF file. In such an implementation of the invention, the formulas would have to be redesigned. In the best mode implementation, before issuing the test application order, the user must identify a CURRENT data period. Under the case-oriented scheme, the user would use a filter to select the desired data DBF file's records, since he would most probably desire to run the tests on a particular set of `readings` (using the terminology just defined in the previous toy factory example). The rest of the application could remain the same, or at least very similar. There would still exist line and group tests (of these, most operations would be group operations). But a new class of group+line tests would have to be defined, in this way: in the best mode implementation there is only one line or record identifier in file tabla.dbf (field LINEA). In the case-oriented scheme there could be more than one. In this way, it would be possible to design formulas that could act on a particular structure of lines. For example, a particular toy could generate not one, but two records for each reading. The user may always want to relate the two lines as if they were one, and thus would like to issue formulas comparing fields in these two records. Similar situations may rise with more than two records, and for different reasons, needs or applications. All these could be solved by the use of more than one record identifier for the tabla.dbf file.
D. Referring to RECURSIVE expert analysis TECHNIQUES: Expert systems are capable of emulating human reasoning through learning the development of multi-layer rule based systems. However, also a simple test-processing scheme could easily emulate these artificial-intelligence techniques in the invention, even before the integration of the expert system within the invention.
In a simple implementation, whenever a particular test turns true, a diagnostic is generated, and the associated classes (and thus actions and triggers) are identified and generated, an action such as a program or program routine may be executed. This program routine may generate a new set of data within the data DBF file tabla.dbf, it may delete sets of data or even modify existing data within the application. In the same manner, tests could be modified, eliminated or added, depending on the class, action and procedure involved. Whenever this happens, the invention would immediately run the tests all over again. This implies the invention may become a self-learning mechanism with recursion, since the loop would stop only when a state of equilibrium is reached. The same procedure could be applied to the summary tests, since the apparition of a particular summary diagnostic may trigger the modification of the summary tests, the original data, or the individual diagnostics on which the summary tests depend. These operations would also, automatically, involve the associated expert system and inference engine.
In another implementation which could be developed by itself or in combination with the previously described implementation, tests could be applied on diagnostics more than once. In the best embodiment, only summary tests are based on previous diagnostics. One summary test is turned on, that is, a summary diagnostic is generated, only when a predetermined set of diagnostics exists. In a multi-layered recursive scheme for summary tests definition, the predetermined set of diagnostics may also include summary diagnostics, and procedure vaplr.prg would be applied on the diagnostics and summary diagnostics repeatedly, until no more summary diagnostics are generated. However, for this procedure to run, a small change needs to be done to vaplr.prg, as described in Section--C: before the first iteration, step 0 would erase all diagnostics for the selected periods before applying the tests; in all other subsequent iterations, no diagnostics would be erased but, instead, all duplicate diagnostics would be erased after applying the test.
In a third possible implementation, summary tests could follow the syntax of procedural logic, or even of the Lisp language and predicate calculus. In the best embodiment, a summary diagnostic is activated whenever a certain number of diagnostics in a predetermined list are true. But requiring all diagnostics to be true is equivalent to an AND list of conditions on all individual diagnostics. In an expanded version, summary diagnostics could respond to combinations of OR, AND and other logical conditioners in procedural logic or in the predicate calculus language.
Expert system building tools such as the VP-EXPERT.TM. expert shell by Paperback Software allow the definition of rules and sets of rules entirely in a database DBF file. These are called induction tables, and are fully documented in available literature. Integrating procedural language techniques into the summary diagnostics is equivalent to developing of an expert system's knowledge base in the invention. In order to convert the invention into a full-fledged expert system, only the inference engine procedures need to be considered. As it was stated previously, the inference engine or rule interpreter has two tasks. First, it examines facts in working memory (the diagnostics that have been turned on) and rules in the rule base (the summary diagnostics), and adds new facts (diagnostics) to the database when possible. Second, it determines in what order rules (summary diagnostics) are scanned and fired. But, the goal in this expert system building tool is to build as many summary diagnostics as possible and, for that goal, the procedure already described, however crude, is an example of a simple but complete forward chaining algorithm; it's not the most efficient, but it serves its task.
These three examples show a few alternative paths that can be followed to turn the invention into a complete expert system.
E. Referring to the invention's integration with EXPERT SYSTEMS and NEURAL NETWORKS: An expert system and inference engine, even a neural network or other kinds of artificial intelligence procedures and techniques, could be integrated with the invention through several procedures. In a simple scheme, the invention would retain its structure and operation as stated in the best embodiment implementation. The inference engine would contain a separate, additional and independent set of rules. Only the data DBF file tabla.dbf, all diagnostics and summary diagnostics and all query operations would be common to both platforms.
In such an implementation, the expert system's inference engine would read its data from the diagnostics DBF file result.dbf, and optionally also from the data DBF file tabla.dbf. It would run its sets of rules on this data, and it would finally generate a set of expert solutions. These solutions would be written as new diagnostics in the diagnostics DBF file result.dbf, using the best embodiment's scheme: a reference (or a list of reference) to the data items involved, and an identifier for the ensuing diagnostic. For compatibility with the invention's current structure, a test DBF file entry would also be generated whenever the corresponding test is missing. The test entry will contain the test's short and long description, and the associated classes, triggers and actions, as described. Thus, the user will be able to also reach the diagnostics generated by the expert system by using the querying engine and the corresponding coordinated diagnostic browse views in the same way described previously. The expert system's diagnostics could be distinguished from all others by using appropriate values for the CLASS1, CLASS2 and CLASS3 fields in file prueba.dbf.
One such possible example of this implementation would involve the VP-EXPERT.TM. expert shell by Paperback Software. In its version 2.0, VP-EXPERT.TM. may read and write DBF database files, accessing them with the GET clause (see sections 5.9 and 9.46 in the VP-EXPERT's user's manual), adding records to them with the APPEND clause (sections 5.19 and 9.8 in the same manual), using DBF files as information base (section 5.3 in the user's manual), etc. Even rules may be created from DBF induction tables (see sections 5.22, 8.4 in the user's manual).
Referring to the use of a DBF file as an information base, VP-EXPERT's manual states in page 5.3: "One of VP-EXPERT's most important and powerful features is its ability to interact with database files created using VP-Info, dBASE II, III, Ill+, or any work-alike database management system." And in page 5.4: "Expert systems frequently require a great amount of data that need period updating. (. . . ) It's far more efficient to store the varying information in a database, and design the rule base to contain a single set of invariant rules. (. . . ) VP-EXPERT's ability to employ this type of interaction is one of its greatest powers. Using the GET clause, data can be retrieved from the fields of a database and assigned as values of knowledge base variables. The PUT and APPEND clauses allow data to be transferred in the opposite direction; they copy values from knowledge base variables to the fields of a database. The MENU clause makes it possible to display menus offering values extracted directly from specified database fields. When a database is used in this fashion, it becomes an information base, and it is considered to be part of the overall knowledge base."
Another important reference is in page 5.21 of said user's manual: "Since VP-EXPERT.TM. permits the user to change data or add data to the knowledge base during a consultation, it can be said that, in a limited sense, the expert system is able to learn from experience. This capacity is significant, because the ability to `learn` is an important criterion in defining an expert system."
The benefits of integrating the invention with existing expert system building tools are: (a) improved efficiency when handling multilayered logic and complex inference procedures and links through multiple indirect addressing, chained statements, and different alternatives and optimizing procedures for the backward or forward chaining logic; (b) user interaction services ideally suited to knowledge databases management, such as the definition of goals, the use of ASK statements to handle missing information, the operation of uncertainty through confidence factors; and finally, (c) tools for facilitating the understanding of the whole logic process. As the number of rules and their complexity grow, it has been shown that it is very hard to manipulate them and it is also hard to fully understand what they do. For this reason, existing expert system building tools incorporate graphic logic trees, trace services, HOW and WHY statements to inquire how the logic proceeds, etc. All these could be eventually integrated into the invention but, given that each particular expert system developing tool in the market today is stronger for a particular set of applications, it's better to allow the integration and add the power of independent inference engines to the invention.
Another artificial intelligence technique is that of NEURAL NETWORKS. Neural networks receive a set of stimuli (diagnostics or data which are defined as the input layer) to produce specific responses (new diagnostics which are defined as the output layer). However, neural networks don't require any kind of programming to determine the relation between stimuli and responses: they learn as they receive a large number of cases where the stimuli and the responses have already been defined. The invention's best embodiment implementation may hold up to sixty periods of data and its associated diagnostics, and the user may, beside all these cases, present the system with many more cases just so it learns. However, the use of a neural network seems to be most appropriate for the alternative configuration described in (C): different data structures for the data DBF file. In the case of the toy factory, or in other quality control applications such as those related to attention to the public, it may be important to determine whether a data record in the data DBF file represent adequate quality or not. These applications are particularly well suited to neural network applications, since sometimes past experience with a large number of cases where it's already well known whether quality was adequate or not may be the best guide to determine what is adequate quality for the future.
The integration scheme of neural networks with the invention is very similar to that of expert systems with the invention: the invention would retain its structure and operation as stated in the best embodiment implementation, and the alternative implementation described in section (C): different data structures for the data DBF file. The data DBF file tabla.dbf, all diagnostics and summary diagnostics and all query operations would be common to the neural system and the invention's best embodiment implementation.
In such an implementation, the neural network would obtain its stimuli from the diagnostics DBF file result.dbf, and optionally also from the data DBF file tabla.dbf. It would run through its input layer, intermediate layer probabilities and the response layer and it would finally generate a set of responses (diagnostics). These diagnostics would be written as new diagnostics in the diagnostics DBF file result.dbf, using the best embodiment's scheme: a reference (or a list of reference) to the data items involved, and an identifier for the ensuing diagnostic. For compatibility with the invention's current structure, a test DBF file entry would also be generated whenever the corresponding test is missing. The test entry will contain the test's short and long description, and the associated classes, triggers and actions, as described. Thus, the user will be able to also reach the diagnostics generated by the neural network by using the querying engine and the corresponding coordinated diagnostic browse views in the same way described previously. The neural network's diagnostics could be distinguished from all others by using appropriate values for the CLASS1, CLASS2 and CLASS3 fields in file prueba.dbf.
Accordingly, the scope of the invention should be determined by the appended claims and their legal equivalents.
Section A--Files of Source Code Given In Appendix A
Section B--Short description of each procedure's operation
Vcambio.prg
Creates and activates main menus. Calls procedures through indirect calls at vmenu.prg and vacti.prg.
Steps
1. Initializes variables
2. Defines menu bars
3. Calls procedures to be executed
Diagnostics menu options are routed through module vacti.prg All other options are routed through module vmenu.prg
vmenu.prg
Calls all program routines from the main menu in vcamblo.prg. Vmenu.prg assigns all menu options and procedures except those under menu option `Diagnostics` from vcamblo.prg to their respective procedures.
vactl.prg
Calls diagnostics menu routines from vcambio.prg
vdrive.prg
Change application path (since each path holds all files for a different application, this is equivalent to choosing an existing or creating a new application). If there are any missing files in the selected directory, vdrive.prg generates these files automatically.
Steps
1. Generate application pick list
2. Accept user selection
3. If User selects empty directory, create all files copying from reference files
4. If directory isn't empty, check files structure against reference files
5. Activate file directory and select file tabla.dbf
vverif.prg
Checks each file's structure by comparing against reference files. That is, vverif.prg verifies the length for each field in each database file. This is run as soon as an application is selected by vdrive.prg.
If there are no reference files to check the application's files against, this same procedure also creates file `struexte` containing information about all fields in each database file, from 1 to 12.
vcond.prg
Add/modify line tests. This module allows the user to modify any existing line tests, or to add new line tests.
vcondg.prg
Add/modify group tests. This module allows the user to modify any existing group tests, or to add new group tests.
vdup.prg
Duplicate a line test.--This module generates a line test by copying an existing line test, so as to allow the user to modify it later. This is good for creating new tests that are very similar to old tests. This option is particularly useful for those line tests that need to be assigned to more than one data item in the data database.
vapl,prg
Applies line tests to all data In selected periods.
Note: In the best embodiment implementation, line tests and group tests are always executed together. Line tests first, and then followed by group tests. After executing vapl.prg, the program will Immediately execute vaplg.prg and vaplr.prg.
vaplg.prg
Applies group tests to all data in selected periods.
veplr.prg
Applies summary definitions and creates summary diagnostics.
This module (module VAPLR) applies summary definitions in files presumen.dbf and dresumen.dbf. This module is executed immediately after the vapl.prg and vaplg.prg modules. Module vaplr.prg reads the diagnostics just generated by vapl.prg and vaplg.prg, and compares them with the summary definitions in files presumen.dbf and dresumen.dbf. It finally writes results to files prresult.dbf and drresult.dbf.
vman.prg
Data browse view.--This module executes the data browse view command option from the Diagnostics menu.
vmanr.prg
Diagnostics browse view.--This module executes the diagnostics browse view command option from the diagnostics menu.
vresu.prg
Frequency browse view.--This module executes the data browse view command option from the diagnostics menu.
vsumm.prg
Summary browse view.--This module executes the data browse view command option from the diagnostics menu.
vbases.prg
File Import--Reads one data period from an ASCII file, and writes it to the desired data period in file tabla.prg,
vlista.prg
Calls the printing routines.
vann.prg
Maintenance--Release one year of information: erases the first twelve months of information and moves all data and related diagnostics twelve months back into the past, thus leaving free space for twelve new months.
Section C--Description of formula application procedure
In the summary description of the intention, formula application procedures are simply referred to by the term procedure Apply.sub.-- formulas.
In the best embodiment Implementation shown in FIG. 84, this procedure becomes three procedures: line tests, group tests and summary tests. These three procedures are always executed one Immediately after the other, and in that order. The procedures involved are: vapl.prg for line tests, vaplg.prg for group tests and vaplr.prg for summary tests.
Important related flies:
Notice that file prueba.dbf (39) contains both line and group tests, but it does NOT contain summary tests, which must be treated separately. Summary tests are defined and stored in files presumen.dbf (45) and dresumen.dbf (46).
Module VAPL.PRG
Uses
Files data.dbf (37), prueba.dbf (39), result.dbf (38).
Steps
0. The user indicates for which period or periods to run the tests. Each period corresponds to a different data fields in the data DBF file (37), tabla.dbf. All tests will only be run for the particular periods indicated by the user. The program will now erase all diagnostics for the selected periods before applying the tests. This ensures there will not be any duplicate diagnostics.
1, identify line tests in prueba.dbf (39): those that contain a data item identifier i.e. a record identifier--field LINEA--, and have the group identifier blank--field GRUPO--
2. operate all line tests in file prueba.dbf (39), one by one in sequence, and for each line test:
a) identify the data items involved in the formula and perform the formula (if-- then condition)
b) if the formula condition is FALSE, goto next test in file prueba.dbf
c) if the formula condition is TRUE, generate diagnostic, that is: write a new record in the diagnostics DBF file (38) result.dbf. This record will only contain the data item identifier (i.e. the line record, field LINEA), the test identifier (field PRUEBA) and the period number Identifier.
Module VAPL.PRG
Uses
Files data.dbf (37), prueba.dbf (39), result.dbf (38), detelle.dbf (41)
Steps
0. The user indicates for which period or periods to run the tests. Each period corresponds to a different data field in the data DBF file, tabla.dbf. All tests will only be run for the particular periods Indicated by the user. The program will now erase all diagnostics for the selected periods before applying the tests. This ensures there will not be any duplicate diagnostics.
1. Identify group tests In prueba.dbf (those that contain a group identifier--field GRUPO--and don't contain a data item identifier, i.e. a. record identifier--field LINEA--).
2. Operate all group tests in file prueba.dbf, one by one In sequence, and for each group test:
a) read the group identifier in file prueba--field GRUPO--
b) search for the next occurrence of that group identifier in file detalle.dbf
c) if this is the end of file detalle.dbf, goto next test in file prueba.dbf
d) if the group identifier was found, now read the corresponding data item Identifier--field LINEA--
e) perform the formula (if--then condition) with the data item pointed to by the value of identifier LINEA just read and the CURRENT period
f) if the formula condition is FALSE, goto (b), I.e., search for next occurrence of that group identifier in file detalle.dbf
g) if the formula condition is TRUE, generate diagnostic, that is: write a new record in the diagnostics DBF file result.dbf. This record will only contain the data item identifier (i.e. the line record, field LINEA), the test identifier (field PRUEBA) and the period number identifier.
Module VAPLR.PRQ
Uses
Files result.dbf (38), presumen.dbf (45), dresumen.dbf (46), prresult.dbf (47), drresult.dbf (48).
Steps
0. The program will first erase all diagnostics for the selected periods before applying the summary tests. This ensures there will not be any duplicate summary diagnostics.
1. Operate all summary tests in file presumen.dbf, one by one in sequence, and for each summary test:
a) delete all records from temporary DBF file memory.dbf
b) read the summary test identifier in file presumen,dbf--field ANALISIS--
c) identify tests; that is:
i) search for the next occurrence in file dresumen.dbf of the summary test identifier just read at file presumen.dbf
ii) if this is the end of file dresumen.dbf, go to (d): check number or occurrences
iii) if the summary test identifier was found, now read the corresponding test identifier (18)--field PRUEBA--
iv) identify all occurrences of the test Identifier (18) in file result.dbf and store all contents of these records in file memory.dbf
v) go to step (i): search for next occurrence of summary test identifier in dresumen.dbf
d) check number of occurrences
i) eliminate all duplicate tests in file memory.dbf. Notice that the criteria for duplicate tests can be defined by the user. This is done simply because the summary test requires a minimum number of non Identical tests in order to be true.
ii) If the number of records in file memory.dbf is smaller than field PMIN for the current summary test in file presumen.dbf, immediately go to next summary test, i.e. go to step (a), delete all records from temporary DBF file memory.dbf
ii) if the number of records in file memory.dbf is greater or equal than field PMIN in file presumen.dbf, then keep going
iii) write a new record in file prresult.dbf, containing the summary test identifier--field ANALISIS-- and the period number identifier--field NPER--
iv) for each one of the occurrences, generate a new record In file drresult.dbf, containing the summary test identifier--field ANALISIS--, the period number identifier--field NPER--, the test identifier (18, field PRUEBA), the data item identifier--field LINEA-- and write code `P` In field POSITIVA.
v) for each test identifier (18) in file drresult.dbf related to the current summary test that doesn't happen in any occurrence in the previous step, generate a new record in file drresult.dbf with the same information, except that field POSITIVA goes blank. This field indicates a TRUE test (`P`) or a test that didn't turn TRUE (blank).
vi) go to next summary test, i.e. go to step (a), delete all records from temporary DBF file memory.dbf
In summary, module vaplr.prg applies summary definitions in files presumen.dbf and dresumen.dbf. For each summary test, it identifies the individual tests that have turned true (i.e. that have associated diagnostics). If these are more than the minimum threshold number PMIN defined In presumen.dbf, then it writes the corresponding records in prresult.dbf and drresult.dbf.
Section D--Description of diagnostic browse operations
Data browse view (vman.prg)
steps
1. Setup screen options
variables initialization 2. Activate files
Select 1: file tabla.dbf (37)
Select 2: file result.dbf (38)
Select 3: file prueba.dbf (39)
3. Define windows in screen
tabla (72, 74, 77 In FIGS. 75, 76, 77)--Upper half of the screen. This will contain selected data from file Select 1, tabla.dbf (37), the data database file (13) result (73 in FIG. 75)--Lower half of the screen. This will contain selected data from file Select 2, result.dbf, the diagnostics database file result1 (75, 78 in FIGS. 76, 77)--May be used instead of `result`. It occupies only 25% of the total screen area, at the lower left corner. When active, contains selected data from file Select 2, result.dbf, the diagnostics database file. At the right of window result1, the program shows related results (76, 79 in FIGS. 76, 77) of prueba.dbf, indexed by the test indicator (i.e., prueba.codigo=result.prueba)
Note: Window `revisar` should be ignored at first when studying the algorithm. This serves mostly cosmetic purposes.
4. Filters selection
Present dialog box to user, for filter selection. Accept user selected filters on fields: linea, prueba, period, clase1,2,3
5. Show first coordinated browses view
Structure:
Cursor control is in WINDOW `tabla` (file TABLA.PFIG).
Window `result` is also active (file RESULT.PRG).
Action:
The cursor keys are browse keys
If the user presses the FS function key, goto step 6.
6. Show second coordinated browses view
Structure:
Cursor control is in WINDOW `result` (file RESULT.PRG).
Window `table` is also active (file TABLA.PRG).
Coordinated diagnostics descriptions (from file PRUEBA.PRG) also shows at the right of window `result1`
Action:
The cursor keys are browse keys
If user presses F5, show first coordinated browses view (step 5)
If user presses F6, show third coordinated browse view (step 7)
The F6 option allows user to find all other data with the same diagnostics.
7. Third coordinated browses view
Structure:
Cursor control is in WINDOW `result1` (file RESULT.PRG).
Window `table` is also active (file TABLA.PRG).
Coordinated diagnostics descriptions (from file PRUEBA.PRG) also show at the fight of window `result1`
Action:
The cursor keys are browse keys
If user presses F6, show second coordinated browse views (step 6)
If user presses F5, show first coordinated browses view and (step 5)
This is a special view that shows in window result1, all occurrences of the diagnostics with the same test identifier codes all throughout the application. That is, before selecting the third coordinated browses view, window result1 showed a few diagnostics (those associated with the data item that was previously selected in window table). After selecting the third coordinated browses view, these same diagnostics still show, but now the result1 window also contains all other occurrences of these diagnostics (those with the same test identifier codes) in all the application. If the user moves the cursor browsing this list of diagnostics, and then presses the F5 key, he will get the corresponding data items in the table window.
General notes
a) F5 and F6 are both switches.
b) If the user presses Escape twice at any time, the procedure ends.
Diagnostics browse view (vmanr.prg)
Steps
1. Setup screen options
variables initialization
2. Activate files
Select 1: file tabla.dbf (37)
Select 2: file result.dbf (38)
Select 3: file prueba.dbf (39)
3. Define windows in screen:
table (80, 82 in FIGS. 78, 79)--Upper half of the screen. This will contain selected data from file select 2, result.dbf, the diagnostics DBF file, and from file select3, prueba.dbf, the tests DBF file. These files are synchronized by the relation result.prueba=prueba.codigo.
result (81, 83 in FIGS. 78, 79)--Lower half of the screen. This will contain selected data from file select 1, tabla.dbf, the data DBF file
Note: Windows revisar and revisar1 should be ignored at first when studying the algorithm. These serve mostly a cosmetic purpose.
4. Filters selection
Present dialog box to user, for filter selection.
Accept user selected filters on fields: lines, prueba, period, clase1,2,3
5. Show first coordinated browses view
Structure:
Cursor control is in WINDOW `result` (file result.prg).
Window `tabla` is also active (file tabla.prg).
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 6.
6. Show second coordinated browses view
Structure:
Cursor control is in WINDOW `table` (file tabla.prg).
Window `result` is also active (file result.prg).
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 5.
General notes
F5 acts like a switch.
If the user presses Escape twice at any time, the procedure ends.
Frequency browse view (vresu.prg)
Steps
1. Setup screen options
variables initialization
2. Activate flies
Select 1: file tabla.dbf (37)
Select 2: file result.dbf (38)
Select 3: file prueba.dbf (39)
Select tamp: file temp.dbf, containing a frequency sum of result.dbf
Select tamp1: file temp1.dbf, containing an extract from result.dbf, the result of filtering all diagnostics by one or more class selections
3. Define windows in screen
tabla (84, 86 in FIGS. 80, 81)--Upper half of the screen. This will contain selected data from file Select tamp: temp.dbf, the temporary diagnostic frequencies DBF file
result (85 in FIG. 80)--Lower hall of the screen. This will contain selected data from file Select tamp1: temp1.dbf, the temporary diagnostics DBF file result1 (87 In FIG. 81)--May be used instead of `result`. It occupies only 25% of the total screen area, at the lower left corner. When active, contains selected data from file Select tamp1:temp1.dbf, the temporary diagnostics DBF file. At the right of window result1, the program shows related results (88 in FIG. 81) of file Select 3: prueba.dbf, indexed by field prueba, the test identifier (18), by the relation prueba.codigo=temp1.prueba.
Note: Window `reviser` should be ignored at first when studying the algorithm. This is created mostly for cosmetic purposes.
4. Filters selection
Present dialog box to user, for filter selection.
Accept user selected filters on fields: linea, prueba, period, clase1,2,3
5. Show first coordinated browses view
Structure:
Cursor control is in WINDOW `tabla` (file TABLA.PFIG).
Window `result` is also active (file RESULT.PRG).
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 6.
6. Show second coordinated browses view
Structure:
Cursor control is in WINDOW `result1` (file temp1.prg)
Coordinated diagnostics descriptions (from file prueba.prg) also shows at the right of window `result1`
Window `table` remains in the screen, but won't refresh its data (file temp.prg).
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 5.
General notes
F5 acts like a switch.
In the best embodiment's code, variable cam represents the condition of the F5 switch.
If the user presses Escape twice at any time, the procedure ends.
Stylized code: The following pseudocode exemplifies simplified operation of this module:
Summary browse view (vsumm.prg)
Steps
1. Setup screen options
variables initialization
2. Activate files
Summary diagnostics DBF file. This is file select 11: prresult.dbf (47). It may be shown in combination with prueba.dbf (39), to show more information related to each summary diagnostic
Summary diagnostics elements DBF file. This is file select 12: drresult.dbf (48).
Summary test DBF file. This is file presumen.dbf (45).
3. Define windows in screen
tabla (89, 92 in FIGS. 82, 83)--Upper half of the screen. This will contain selected data from file Select 11, prresult.dbf, the summary diagnostics definitions DBF file, indexed by presult.prueba=prueba.codigo to file prueba.dbf, the tests DBF file.
result (90, 93 in FIGS. 82, 83)--Lower half of the screen. This will contain selected data from file Select 12, drresult.dbf, the summary diagnostics elements DBF file
The middle section of the screen (between windows tabla and result) is not a window, but the program will display the diagnostics descriptions there (91, 94 in FIGS. 82, 83).
Note: Window `revisar` should be ignored when studying the algorithm. This is created for cosmetic purposes.
4. Show first coordinated browses view
Structure:
Cursor control 1s in WINDOW `table` (files select 11: prresult.dbf end presumen.dbf).
Window `result` is also active (files select 12: drresult.dbf and prueba.dbf).
Description window at the middle of the screen, dynamically moving indexed by prresult.analisis=presumen.analisls with cursor control (file: presumen.dbf)
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 6.
5. Show second coordinated browses view
Structure;
Cursor control is in WINDOW `result` (files select 12: drresult.dbf and prueba.dbf).
Window `tabla` is also active, but immobile (files select 11: prresult.dbf and presumen.dbf).
Description window at the middle of the screen, dynamically moving Indexed by drresult.prueba=prueba.codigo with cursor control (file: presumen.dbf)
Action:
The cursor keys are browse keys
If the user presses the F5 function key, goto step 5. ##SPC1##
Claims
What is claimed:
1. An apparatus for analyzing an array of data stored in a quantitative database, comprising: a memory for storing said data of said quantitative database as a table in a relational database; a computer coupled to said relational database and having a display and having an input device for receiving user input, said computer for execution of one or more programs so as to process data from said quantitative database; a first program in execution on said computer for controlling said computer to receive user input defining one or more analysis rules to be performed on a subset of data, also specified by said user input, from said array of data stored in said quantitative database and for controlling said computer so as to analyze selected ones of said user specified subset of data items in accordance with selected ones of said one or more user specified analysis rules so as to generate one or more diagnostic records the text of which is also defined by said user input, each diagnostic record corresponding to a possible result of the application of a particular user specified analysis rule to the corresponding user selected subset of items of data from said array of data in said quantitative database, and wherein said first computer program controls said computer so as to automatically generate one or more link pointers for each said diagnostic record, each said link pointer linking a diagnostic record to the corresponding user selected subset of items of data from said quantitative database from which said diagnostic record was generated; and a second program in execution on said computer for controlling said computer so as to store each said diagnostic record in a diagnostic database in said memory and wherein said link pointer associated with each diagnostic record is programmable by said user, and wherein said first program includes one or more routines for controlling said computer to display at least first and second windows on said display, and for controlling said computer to display in said first window at least some of said items of data from said quantitative database and for controlling said computer to display in said second window at least some of said diagnostic records, and further comprising one or more routines in said first computer program for controlling said computer to display a pointer in said second window and to receive and process user data controlling the position in said second window of said pointer such that any diagnostic record displayed in said second window may be selected by said user using said pointer, and when any said displayed diagnostic record in said second window is so selected, for controlling said computer to display in said first window the corresponding subset of data items from said quantitative database associated by one or more of said link pointers with the selected diagnostic record.
2. The apparatus of claim 1 wherein said first program controls said computer to receive user input in the form of one or more programmable text strings to be stored in said memory with each user supplied alphanumeric text string corresponding to a possible result stemming from application of a particular analysis rule, each said programmable text string being a diagnostic statement which will be stored as one of said diagnostic records in said database if the result corresponding to said diagnostic statement occurs upon execution of the corresponding analysis rule.
3. The apparatus of claim 1 wherein each said analysis rule has only two possible results, true or false, which may be returned after application of said analysis rule to the corresponding user specified data from said quantitative database, and wherein said second computer program further comprises a routine for displaying on said display the diagnostic record(s) resulting from application of selected ones of said user specified analysis rules which returns a true result before, during or after recording said diagnostic record in said memory.
4. The apparatus of claim 1 wherein said second program includes a routine to store all said link pointers generated by said second program in carrying out processing defined by the user selected ones of said analysis rules to be executed, the resulting link pointer or pointers being stored in said memory, each said link pointer relating a particular user selectable subset of data in said quantitative database to a particular corresponding diagnostic record.
5. The apparatus of claim 1 wherein said first program includes a routine for receiving and processing user input in the form of commands to program each said analysis rule as a conditional branching test wherein a user selected subset of items of data from said quantitative database is compared to one or more test criteria using mathematical or logical operators or both and for receiving and processing commands entered by the user to define at least two transfer control branches for a particular user selected analysis rule to control transfer of processing by said computer to at least two different points in said first program depending upon the result of said comparison performed by said analysis rule, and wherein, for each said branch, a different diagnostic statement is displayed and stored in said memory as a diagnostic record.
6. The apparatus of claim 1 wherein said first program includes one or more routines for receiving and executing user input which defines a set of one or more expert test analysis rules each of which has at least two possible results, each of said expert test analysis rules for analyzing one or more user selectable diagnostic records from said diagnostic record database stored in said memory, each of said expert test analysis rules having a user defined diagnostic statement associated with at least one of said two or more possible results, each of said expert test analysis rules being an If-Then statement implementing a test condition wherein one or more of said diagnostic records are tested for satisfaction of a particular, user defined logical or mathematical relationship, and if the test condition is satisfied corresponding to a particular one of said two or more possible results, for entering a new diagnostic record in said diagnostic database comprised of the diagnostic statement associated with the test condition result which occurred.
7. The apparatus of claim 1 wherein said first computer program includes one or more routines which control said computer to process and execute user input which defines the link pointer associated with each said diagnostic record, and wherein said first program includes one or more programs for controlling said computer to display on said display at least a first window in which there is displayed a subset of data items from said quantitative database and a second window in which there is displayed at least one diagnostic record from said diagnostic database, and further comprising a query program in execution on said computer for controlling said computer to display on said display a pointer in said first window and for controlling said computer to receive and process user input data by which said pointer in said first window may be manipulated by said user to select a subset of data items displayed in said first window, and for controlling said computer to use any said link pointers associating the selected subset of data items to one or more of said diagnostic records for retrieving from said diagnostic database and displaying in said second window the diagnostic records associated with said selected subset of data items displayed in said first window.
8. An apparatus for analyzing an array of data stored in a quantitative database, comprising: a memory for storing said quantitative database data as a table in a relational database; a computer coupled to said relational database and having a display and having an input device for receiving user input, said computer for execution of one or more programs so as to process data from said quantitative database; a first program means for execution by said computer for controlling said computer to receive user input defining one or more analysis rules to be performed on one or more subsets of data items, also specified by said user input, from said array of data stored in said quantitative database] and for controlling said computer so as to analyze selected ones of said subsets of user specified data items in accordance with selected ones of said one or more user specified analysis rules so as to generate one or more diagnostic records the text of which is also defined by said user input, each diagnostic record corresponding to the application of a particular user specified analysis rule to the corresponding user selected subset of data items from said array of data in said quantitative database, and wherein said first computer program means includes means for controlling said computer so as to automatically generate one or more link pointers for each said diagnostic record, each said link pointer linking a diagnostic record to the corresponding user selected subset of data items from said quantitative database from which said diagnostic record was generated using the associated analysis rule; and a second program means for execution by said computer for controlling said computer so as to display on said display all diagnostic records resulting from carrying out said analysis rules selected by said user on the user selected subset of data items from the quantitative database and for controlling said computer to store said link pointers in an index pointer table in said memory, and for storing each said diagnostic record in a diagnostic database in said memory; and wherein said first program means includes means for controlling said computer to receive and process user input data which defines the linkage between any particular analysis rule and the subset of data items from said quantitative database upon which said analysis rule is to be executed, said linkage being embodied in link definition data which is stored with the analysis rule defined by the user and which defines how said computer later generates a link pointer for any diagnostic record which resulted upon return by an analysis rule of a predetermined result and which defines which subset of data from said quantitative database is linked to said particular diagnostic record, and wherein said first program means includes means for controlling said computer to display at least first and second windows on said display, and for controlling said computer to display in said first window at least some of said subsets of data items from said quantitative database and for controlling said computer to display in said second window at least some of said diagnostic records, and further comprising means in said first computer program for controlling said computer to display a pointer in said second window and to receive and process user data controlling the position in said second window of said pointer such that any diagnostic record displayed in said second window may be selected by said user using said pointer, and when any said displayed diagnostic record in said second window is so selected, for controlling said computer to display in said first window the corresponding subset of data items from said quantitative database associated by one or more of said link pointers with the selected diagnostic record.
Patent Citations (29)
| Patent | Date | Inventor | Cited By |
|---|---|---|---|
| US4658370 | 1987-04-01 | Erman et al. | |
| US4754409 | 1988-06-01 | Ashford et al. | |
| US4847784 | 1989-07-01 | Clancey | |
| US4866635 | 1989-09-01 | Kahn et al. | |
| US4884218 | 1989-11-01 | Agnew et al. | |
| US4891766 | 1990-01-01 | Derr et al. | |
| US4920499 | 1990-04-01 | Skeirik | |
| US4928236 | 1990-05-01 | Tanaka et al. | |
| US4945476 | 1990-07-01 | Bodick et al. | |
| US4949278 | 1990-08-01 | Davies et al. | |
| US4974196 | 1990-11-01 | Iwami et al. | |
| US4982344 | 1991-01-01 | Jordan | |
| US5006992 | 1991-04-01 | Skerik | |
| US5025392 | 1991-06-01 | Singh | |
| US5034898 | 1991-07-01 | Lu et al. | |
| US5043915 | 1991-08-01 | Suwa et al. | |
| US5099436 | 1992-03-01 | McCown et al. | |
| US5103498 | 1992-04-01 | Lanier et al. | |
| US5121496 | 1992-06-01 | Harper | |
| US5127005 | 1992-06-01 | Oda et al. | |
| US5131087 | 1992-07-01 | Warr | |
| US5159669 | 1992-10-01 | Trigg et al. | |
| US5164912 | 1992-11-01 | Osborne et al. | |
| US5167012 | 1992-11-01 | Hayes et al. | |
| US5193143 | 1993-09-01 | Kremmer et al. | |
| US5204974 | 1993-04-01 | Bernstein et al. | |
| US5214653 | 1993-05-01 | Elliott, Jr. et al. | |
| US5228116 | 1993-07-01 | Harris et al. | |
| US5257185 | 1993-10-01 | Farley et al. |
Non-Patent Literature (1)
- Timeworks, "Swift Call St User's Manual", 1982, pp. 93-97, 108-109.