US 5,900,870 AGrant

Object-Oriented Computer User Interface

Issue Date:1999-05-04

•35 Claims

•28 Drawing Sheets

Abstract

A computer user interface includes a mechanism for graphically representing and displaying user-definable objects of multiple types. The object types that can be represented include data records, not limited to a particular kind of data, and agents. An agent processes information automatically on behalf of the user. Another mechanism allows a user to define objects, for example by using a template. These two mechanisms act together to allow each object to be displayed to the user and acted upon by the user in a uniform way regardless of type. For example, templates for defining objects allow a specification to be input by a user defining processing that can be performed by an agent.

Metadata

Assignee

Massachusetts Institute of Technology

Inventors

Thomas W. Malone
Kum-Yew Lai
Keh-Chiang Yu
Richard W. Berenson

Application Information

Application Number:US 3367126

Filing Date:1994-11-09

Priority Date:1989-06-30

Art Unit:272

Classifications

IPC:

G06F 1500

Field of Search:

395345155;156;157;158;159;160;161333;334;335;345;356;357

Patent Drawings (28 sheets)

Description

Technical Field

Background Art

Summary of the Invention

Brief Description of the Drawings

Detailed Description of Specific Embodiments

1. KEY IDEAS

Semistructured Objects

Template-based user interfaces

Relationships among objects

Tailorable display formats

Inheritance hierarchy for objects

Customizable Folders

Semiautonomous Agents

Descriptions

2. SYSTEM FEATURES

Terminology: Objects and templates

Editing Instances

Creating New Instances

Creating New Object Types

Changing the Display Format and Other Properties of Object Types

Folders

Performing Actions on Objects

Object specific actions

Context specific actions

Displaying and invoking actions

Creating Agents and Rules

Embedded descriptions

Navigating Through the System

Saving and Sharing Knowledge

Saving knowledge

Sharing knowledge by sending messages

Sharing knowledge by transferring files

Object Lens as a Network User Operating Environment

3. OTHER APPLICATIONS

Task Tracking

Intelligent Message Sorting: Engineering Change Notices

Database Retrieval

Hypertext

An Example: Representing the Structure of Arguments

Step 1: "Simple" Hypertext

Step 2: Creating Individual Nodes

Step 3: Summarizing the Contents and Relationships in Groups of Objects

Step 4: Automatically Selecting and Manipulating Objects

Specialized Display Formats: Calendar

Automatic Agents for Searching and Manipulating Networks

4. SYSTEM ARCHITECTURE

What is "Object-Oriented"

Object Manager

Forms Manager

Agent Manager

Mail Manager

Database

Lens Object Classes

Form

System

Thing

Message

Rule

Folder

Types Folder or "View" Folder

Agents

Application Document

Type

Description

Variables

Actions

Commands

Detailed Technical Documentation

Organization of the LOOPS Classes in Object Lens

Methods by Class

Editor Mixin

Methods for Presenting a Template-Form

Methods for Manipulating the contents of a Template-form

Methods for Updating an Object from the Contents of a Template-Form

Methods which Only Affect the Saved Object

TableMixin

TreeMixin

Lens Object

BrowserMixin

Executive

Action

Action's Progeny

Template (Type)

Thing

Message

Folder

Agent

Rule

Single Instance Classes

The Organization of the File

Additional Requirements

1. Technical Field

The present invention relates to computer systems generally, and specifically to the portions of computer systems designed to display and to make available to the user the information stored therein.

2. Background Art

With the increasing power of microprocessors, and of computers generally of any given physical size, there has been a widely recognized need for systems that would permit users who lack sophisticated programming skills to utilize this newly available computational power for a wide range of tasks. Many different approaches have been taken in order to try to satisfy this need.

One line of approach has tried to tame the ordinary and often complicated user interface of a computer system's operating system by providing an "operating environment" within which the user relates to the operating system without directly using the operating system's commands. The Macintosh Finder is an example of this approach.

Another approach is described in a recent paper by T. W. Malone et al., entitled "Semistructured Messages Are Surprisingly Useful for Computer-Supported Coordination," 5 ACM Transactions on Office Information Systems, No. 2, April 1987, pp. 115-131. In that paper, there are introduced the concepts of "semi-structured messages" and of "rules" for processing of these messages that are used in a system called "Information Lens".

The approach of operating environments is limited in that they are dependant on specific applications programs to provide access to databases having wide ranges of applicability to users, and the specific applications are not necessarily integrated with the operating environment. Furthermore, although providing some valuable concepts, Information Lens is limited to facilitating electronic mail communications and is not applicable to general databases.

SUMMARY OF THE INVENTION

The present invention provides a graphical interface between a user and a database (which may be shared among multiple users who may access the database at the same or different times) in such a way that the user may functionally access and modify the contents thereof for a wide range of applications in a uniform, intuitive, and simple manner. The invention harnesses the power of concepts from "objected oriented programming" for the benefit of computer users who are not sophisticated programmers. In a preferred embodiment, the present invention provides a computer user interface that includes an arrangement for the graphic representation of data records and agents as instances of types of objects. Each object type is defined by a template and has its own semistructured format for presenting data pertaining to instances of the object type. In a further preferred embodiment, messages, rules, and folders are also pre-defined types of objects, and object types may be defined as part of an inheritance hierarchy.

The terms "data record", "agent", "object", "type" of object, "instance" of object, "template", "folder", "rule", "semistructured", and similar terms have the same meanings in the claims set forth below as they do in the following description.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates, in a preferred embodiment of the invention, how objects can be edited with a simple template editor.

FIG. 2 illustrates, in a preferred embodiment the use of embedded templates to allow related objects to be viewed and edited simultaneously.

FIG. 3 illustrates, in a preferred embodiment, an example of how object types may be defined hierarchically.

FIG. 4 illustrates, in a preferred embodiment, the presentation of an instance of an agent.

FIG. 5 illustrates, in a preferred embodiment, an instance of a rule.

FIG. 6 illustrates an instance of a rule using embedded descriptions to create complex queries.

FIGS. 7a and 7b and 7c illustrate examples of reports created by presenting a folder through tables and trees as provided by a preferred embodiment.

FIG. 8 illustrates how a folder presented as a table can be used to summarize selected fields from Action Request messages.

FIG. 9 illustrates how rules can include multiple levels of embedded descriptions that refer to linked objects throughout the knowledge base.

FIG. 10 illustrates how a rule can be used to formulate a complex query to retrieve all the objects from a database that satisfy certain criteria.

FIG. 11 illustrates, in a preferred embodiment, how an object may be a hypertext document that can include links, not only to other text passages, but also to other object types such as people and bibliographic citations.

FIG. 12 illustrates, in a preferred embodiment, another hypertext example.

FIG. 13 illustrates, in the context of the systems of FIG. 12, an argument node.

FIGS. 14(a) and (b) illustrate tree and table display formats in the same context.

FIG. 15 illustrates the ability to select objects from another folder to be inserted.

FIG. 16 illustrates the use in this context of an agent to notify the user when arguments have been added.

FIG. 17 illustrates the rule used by the agent of FIG. 16.

FIG. 18 illustrates in the same context a multiply embedded rule that may be employed to use information throughout a user's knowledge base.

FIG. 19 included FIGS. 19A-19C illustrates a specialized display format called "Calendar" used in a preferred embodiment of the invention.

FIG. 20 illustrates how a rule-based agent may perform automatic actions such as maintaining a folder of overdue tasks.

FIG. 21 included FIGS. 21A-21C presents models of user operating environments in accordance with the prior art and with a preferred embodiment of the invention.

FIG. 22 illustrates object oriented programming object classes in accordance with a preferred embodiment of the invention.

FIG. 23 illustrates the structure of a form in accordance with a preferred embodiment of the invention.

FIG. 24 illustrates an example of template form data in accordance with a preferred embodiment of the invention.

FIG. 25 shows the modular architecture of a preferred embodiment of the invention.

FIG. 26 shows the object class and methods structure, at the program level, of the object lens system, in accordance with a preferred embodiment of the invention.

DETAILED DESCRIPTION OF SPECIFIC EMBODIMENTS

The present invention in a preferred embodiment has been implemented as a prototype in Interlisp-D on Xerox 1100 series workstations connected by an Ethernet. The prototype embodiment is called Object Lens, and it makes heavy use of the object-oriented programming environment provided by LOOPS, the built-in text editor, Tedit, which are available for use in conjunction with Interlisp-D. The source code listing for Object Lens is attached hereto and incorporated herein by reference.

The following description of Object Lens begins with a presentation of its key ideas. Thereafter are discussed features of the system with some simple examples, followed by a discussion of how these features can be combined to create a variety of cooperative work applications. Next is discussed the system architecture of Object Lens, and finally the specific program structure.

1. KEY IDEAS

One of the most important characteristics of Object Lens is that it is a semiformal system. We define a semiformal system as a computer system that has the following three properties: (1) it represents and automatically processes certain information in formally specified ways; (2) it represents and makes it easy for humans to process the same or other information in ways that are not formally specified; and (3) it allows the boundary between formal processing by computers and informal processing by people to be easily changed.

Semiformal systems are most useful when we understand enough to formalize in a computer system some, but not all, of the knowledge relevant to acting in a given situation. Such systems are often useful in supporting individual work, and we believe they are especially important in supporting cooperative work where there are usually some well-understood patterns in people's behavior, but where there is usually also a very large amount of other knowledge that is potentially relevant but difficult to specify.

In order to create such a flexible semiformal system, the knowledge embodied in the system must be exposed to users in a way that is both visible and changeable (cf., Turbak, 1986). That is, users must be able to easily see and change the information and the processing rules included in the system. In Object Lens, there are three key ideas about how to represent and expose knowledge to users:

(1) "Passive" information is represented in semistructured objects with template-based interfaces;

(2) "Aggregate" information from collections of objects is summarized in customizable folders; and

(3) "Active" rules for processing information are represented in semiautonomous agents.

Discussed below is how semistructured objects and semiautonomous agents allow us to expose knowledge to users in a way that is both visible and changeable.

Semistructured Objects

Users of the Object Lens system can create, modify, retrieve, and display objects that represent many physically or conceptually familiar things such as messages, people, meetings, tasks, manufactured parts, and software bugs. The system provides an interface to an object-oriented database in the sense that (1) each object includes a collection of fields and field values, (2) each object type has a set of actions that can be performed upon it, and (3) the objects are arranged in a hierarchy of increasingly specialized types with each object type "inheriting" fields, actions, and other properties from its "parents" (see Dittrich & Dayal, 1986; Shriver & Wegner, 1987; Stefik & Bobrow, 1986). For example, a TASK object may have fields like Requestor, Performer, Description, and Deadline; a PERSON object may have fields like Name, Phone, Address, and Job title; and a STUDENT object may add fields like Year and Advisor to the fields present in all PERSON objects. Some objects (e.g., MESSAGES) have specialized actions defined for them (e.g., Answer and Forward). As described in more detail below, we have provided rudimentary facilities for saving and sharing objects, and we are currently exploring ways to link our interface to remote databases.

The objects in Object Lens, like messages in the Information Lens, are semistructured in the sense that users can fill in as much or as little information in different fields as they desire and the information in a field is not necessarily of any specific type (e.g., it may be free text, such as "I don't know").

Template-based user interfaces. Users can see and change objects through a particularly natural form of template-based user interface. These interfaces have a number of virtues. For instance: (1) they resemble forms, with which users are already familiar, (2) they conveniently inform users about the fields contained in an object and about other information such as the likely alternatives for different fields, and (3) their use is consistent across many different kinds of objects. We will see below how this interface approach, which was used for messages and rules in the Information Lens, can be easily generalized to many different kinds of objects.

Relationships among objects. Users can easily see and change the relationships among objects by inserting and deleting links between the objects. For instance, the Requestor and Performer fields of a Task object might contain links to the Person objects that represent, respectively, the person who will perform the task. Then, for instance, when the user looks at the Task object, it will be easy to get more information (e.g., the phone numbers) about the people involved with the task. We will see below how this capability of linking objects to each other provides a rudimentary hypertext system as a special case (see Conklin, 1987, for an extensive review of hypertext systems). We will see below how it is also possible for an object to which a link appears to be displayed as an embedded template inside the original template.

Tailorable display formats. Users have several options for changing the ways they see objects. For instance, they can easily: (1) select which fields will be shown and which will be suppressed, (2) rename selected fields, and (3) specify the default and alternative values the system presents for individual fields.

Inheritance hierarchy for objects. The creation and modification of type definitions is simplified by arranging object types in an inheritance hierarchy (e.g., Stefik & Bobrow, 1986). New types of objects are defined as specializations of existing object types, and they automatically "inherit" all properties of the existing objects except those which are specifically "overridden." Since most of the information about new object types can thus be "inherited" from existing types, rather than having to be re-entered each time, creating new object types becomes simpler. Also, when an object type definition is changed later, the changes are automatically "inherited" by the specializations of that object type.

Customizable Folders

Users of Object Lens can group collections of objects together into special kinds of objects called Folders. For instance, folders can be created for groups of people (e.g., project teams, company directory), tasks (e.g., those completed, those to be done by you, those to be done by others), messages (grouped according to topic or urgency), and so forth. Users can also easily customize their own displays to summarize the contents of objects in a folder. For instance, they can select certain fields to be displayed in a table with each row representing an object in the folder and each column representing a field. They can also select fields from which the links between objects will be used to create a tree (or graph) display with each object represented as a node in the tree and each link in the selected field represented as a line between nodes.

Semiautonomous Agents

Users of the Object Lens system can create rule-based "agents" that provide specifications for processing information automatically on behalf of their users. (It is, however, within the scope of the invention to utilize processing specifications that are other than rule-based.) These agents provide a natural way of partitioning the tasks performed automatically by the system. As we will see below, agents can be "triggered" by events such as the arrival of new mail, the appearance of a new object in a specified folder, the arrival of a pre-specified time, or an explicit selection by the user. When an agent is triggered it applies a set of rules to a specified collection of objects. If an object satisfies the criteria specified in a rule, the rule performs some specified action. These actions can be general actions such as retrieving, classifying, mailing, and deleting objects or object-specific actions such as loading files or adding events to a calendar.

The agents in Object Lens are "autonomous" in the sense that once they have been created, they can take actions without the explicit attention of a human user. They are only "semiautonomous," however, in the sense that (a) they are always controlled by a human user (that is, all their rules can be easily seen and changed by their human user), and (b) they may often "refer" objects to their human user for action (e.g., by leaving the object in the user's in box) rather than taking any actions on their own.

Descriptions. Since agents and rules are themselves objects, users can see and modify them with the same template-based user interface that is used for all other kinds of objects. To specify the criteria for when a rule should act upon a given object, users create descriptions of the objects to which the rules apply. A description is simply a partially filled-in template for an object of a particular type. Descriptions can also include embedded descriptions that specify characteristics that must be satisfied by objects to which the original object is linked. For instance, a description of a Task might include an embedded description of the Person who will perform the task. These embedded descriptions (like those in the Rabbit system Tou et al., 1982!), allow users to easily specify object retrieval operations that are equivalent to "joins" followed by "selects" in a relational database.

2. SYSTEM FEATURES

Terminology: Objects and templates. In this section, we will describe in more detail the basic system features of Object Lens and illustrate them with simple examples. Before proceeding it is helpful to clarify some terminology concerning objects and templates. First, we distinguish between object types (or "classes") and specific object instances (e.g., see Fikes & Kehler, 1985). We use the term object type to refer to a kind of object (such as Person or Task) and the term object instance (or simply "instance") to refer to a specific example of one of these object types (e.g., "Joe Smith" or "Task No. 17"). In contexts where the distinction between object types and object instances is not critical, we use the term objects to include both.

We also use the term template in two ways. First, in a general sense, in this description and in the following claims, we use the term template to mean a graphical display of field names and contents in consistent relative locations. Most of a user's interactions with Object Lens are based on such templates. Second, in the Object Lens screen displays, we use the word Template to mean object type definition. (When we use Template in this specialized sense, we will always capitalize it. In some places we also use the word "Type" interchangeably with "Template".) For instance, users can change the display format for all Person objects by editing the Template that defines the Person object type.

Editing Instances

FIG. 1 shows a template for an instance of a person. Using the built-in text editor, users can insert text or bitmaps in any field. In addition, when users click on a field name with the mouse, a list of likely alternative values for that field appears in a pop-up menu. The alternatives may be links to other objects or just text strings. Selecting one of these alternatives causes the alternative to be automatically inserted in the field. For instance, the figure contains a link to the Person object representing Kum-Yew Lai's supervisor. To insert links to objects that are not in the alternative list, the user (a) positions the cursor at the place in the template where the link is to be inserted, (b) selects the Add Link option from the menu at the top of the window, and (c) points to the object to which the link should be made. After a link is inserted, clicking on it with the mouse causes the object it points to appear on the screen.

In the current version of Object Lens, users can insert any combination of text, numbers, links, and bitmaps in any field. Then, in some cases, type checking is done when the editing window for the instance is closed or when certain kinds of processing are done. For instance, the To and CC fields are checked for valid folders (see below for descriptions of rules and folders). In future versions of Object Lens, we will experiment with more restrictive type enforcement in certain fields. For instance, it should probably be impossible to even insert something other than a folder in the "move to" field of a rule action.

FIG. 2 shows a slightly more complex template; this one is for a Bug Fix Request message. One of the fields of this template is the Bug to be fixed and the value of this field is a link to a Bug object. In this case, instead of simply showing a link to the Bug object, the template contains an embedded template for the Bug object itself. The fields in this embedded template can be edited just like the rest of the fields in the template. We will see below how users can specify whether links to other objects should be displayed as link icons (as in FIG. 1) or as embedded templates (as in FIG. 2).

Creating New Instances

To create and display a new instance of an object type that already exists, users click with the mouse on the definition (i.e., the Template) for that object type. FIG. 3 shows the Templates currently included in our system. For instance, to send a new message, users click on the Template for the type of message they want to create; to create a new person object, users click on the Person Template. Then an object instance, like those shown in FIGS. 1 and 2, will appear and the user can fill it in.

Creating New Object Types

To create a new object type, users click (with both mouse buttons, instead of the left one) on the Icon for the "parent" object type (see FIG. 3). This causes a menu to appear showing alternative actions that can be performed on a Type. One of these actions is to Create a subtype. When the user selects this action, a new type is created with all the fields and properties of its "parent." Then users can add fields to the new Type or change its display format and other properties (see below).

In the current version of Object Lens, all Things have three fields: Name, Keywords, and Comments. All objects inherit these fields, though as discussed below, some objects rename these fields or suppress their display. For instance, Messages rename the Name Field to be Subject and the Comments field to be Text.

Changing the Display Format and Other Properties of Object Types

To change the display format or other properties of an object type, users "edit" the Template that defines the object type. Users make these changes by selecting actions from the menu that appears when they click on the Type Icon (as shown in FIG. 3) with both mouse buttons. In this way, users can change (a) which fields of the Type's instances are actually displayed, (b) the names that are displayed for those fields, (c) the alternative values that are displayed for each field, (d) the default values that are displayed in each field when new instances are created, and (e) whether the links in a field should be shown as link icons (see FIG. 1) or as embedded templates (see FIG. 2). Users can also add or delete fields from a Type. All the changes made to a Type are applied to old instances of an object type as well as to newly created ones. For example, if the user changes the name of a field, then the new name will be shown when any old instances are redisplayed.

We anticipate that this system will be used with a core set of object types shared by the users in a group and that the fields in these types will be modified only by an "authorized view administrator." Other users will be able to change the display format of these types (e.g., suppress the display of a field or change its name), but they would not be able to delete or add fields to these "official" types. All users would, however, be able to create their own types as specializations of the official types, and for these types they could add and delete new fields as desired. Elsewhere (Lee & Malone, 1988) we have proposed a scheme for letting an arbitrarily large number of groups share partially overlapping sets of type definitions in arbitrary ways. One of the key ideas of this scheme is that specialized types created by one group can be interpreted by members of another group as instances of the most specific "ancestor" type that both groups share. For instance, a "Student" object created by one group might be interpreted as a "Person" object by another group that does not have a definition for "Student."

Folders

As noted above, Object Lens users can group collections of objects together into special kinds of objects called Folders (see FIG. 7). An object can be added to a folder in two ways: (1) automatically, as the result of a rule action, or (2) manually, using the Add Link action from the *Others* submenu on the folder. In both cases, the folders will contain links to the objects, not the objects themselves. Therefore, the same object can appear in more than one folder. Other actions for moving, copying, and deleting both objects and links are described below.

Object Lens currently provides two formats for displaying the contents of folders: tables and trees. Tables show the values of selected fields from the objects contained in the folder. For instance, FIG. 7a shows a folder that contains objects representing people with the fields displayed for a simple office directory. Users can easily tailor the format of these displays by selecting from a menu the fields they want to have included in the table. For instance, FIG. 7b shows the same folder, but with the display format changed to include a different set of fields.

Trees are graphs that show the objects in a folder and the links that connect these objects. Just as users can select the fields to be shown in a table, they can also select the fields from which links will be shown. For instance, FIG. 7c shows the same folder again, but this time in tree format with the "Supervisor" field selected as the one from which links are displayed. In this case, the display resembles a simple organization chart. In the current version of Object Lens, links from multiple fields can be shown with the links from different fields being displayed as different types of lines (e.g., solid, dotted, etc.). This format may be described generally as a node-arc graph.

When a new folder is created, the user is asked to select the default object type to be contained in the folder. The user is then allowed to choose from the fields of this default object type when selecting the fields to show in a table or when selecting the fields from which links will be shown in a tree. Even though all folders have default object types, no strict type checking is enforced. If an object of an unexpected type is inserted into a folder, only the fields it shares with the default type are displayed in tables and trees.

Performing Actions on Objects

In addition to editing the contents of objects, users can also perform pre-defined actions on them. The actions that can be performed at any time depend on two primary factors: (1) the type of object being acted upon, and (2) the context in which the action is invoked.

Object specific actions. Each object type has a set of actions that can be performed on it. Some of these actions are "inherited" directly from the "parents" of the object type. Others may be modified or added specifically for this object type. For instance, there are some actions, such as Hardcopy and Save that can be performed on all objects (i.e., all instances of Thing and all its subtypes). (Some of these actions, such as Hardcopy, are not yet implemented for all object types.) In addition, more specialized types of objects have other actions defined for them. For instance, agents have a Trigger action that triggers them to start running, and folders have a Change Display Format action that changes them from table format to tree format or vice versa.

In a few cases, the object specific actions depend not just on the type of object, but also on its state. For instance, messages created on the local workstation have a Send action, and messages received from elsewhere have actions such as Answer and Forward. So far these state-specific actions on objects are implemented as special cases. However, we are experimenting with a more general mechanism for representing state-specific actions and making this representation accessible to users. This mechanism defines each action as an object which contains a procedural description written in an Object Action Language. These user-defined actions are treated as object specific actions, and are constructed from the primitive actions available in the language. For instance, a user might create a modified "Hardcopy" action that does a hardcopy only if the "Printable" field of the objects is "True".

Context specific actions. There are some actions that can be applied to any kind of object, but which can be invoked only from certain contexts. The primary contexts are: (1) from a template (like the one in FIG. 1), (2) from a folder that contains the object, (3) from a rule operating on the object, and (4) from a link icon for the object.

For instance, when an object is being displayed in a template, there are several kinds of actions, such as Close, Move, and Shape, that apply to the editing window. Other actions in a template include: (a) Add Link, (insert at the current cursor position a link to another object selected by the user), and (b) Cancel (close the window without saving any of the changes made since the window was last opened.)

When an object is displayed in a folder, other context-specific actions can be applied to it, such as: (a) Show (open an editor on the object), and (b) Select (select the item for some later folder action such as Delete Selection).

The actions that can be applied to an object by rules are discussed below. The actions that can be applied to link icons include: Show (open an editor on the object), and Delete (delete this link to the object).

Displaying and invoking actions. Users invoke the above actions in slightly different ways depending on the context in which the object is displayed. If the object is displayed in a template (like the one in FIG. 1), then several of its most common actions are shown across the top of the template, and all other actions are shown in a menu that pops up when the *Others* action is selected.

When a link to an object is displayed (either as a link icon or as a row in a table), users can invoke actions in two ways. First, if users click on the link with the middle mouse button, a menu pops up showing all possible actions on the object. In addition, simply clicking on the link with the left mouse button invokes the most common action. For instance, clicking with the left button on a row in a table Selects the object for subsequent folder actions, while clicking with the left button on a link icon inside an editor Shows the object in another window on the screen.

Creating Agents and Rules

In some cases, agents can take actions automatically on behalf of their users. For instance, FIG. 4 shows an example of a simple agent designed to help a user process incoming mail. When an agent is triggered, it applies a set of rules to a collection of objects in a folder. The agent in FIG. 4 is applied to objects in the New Mail folder and is triggered by the arrival of new mail. That is, when mail is retrieved to the workstation, the mail program automatically inserts links to the new messages into the user's New Mail folder and these New Links trigger the agent. In the current version of Object Lens, two other kinds of automatic triggers are available: Daily at Midnight, and On the Hour.

The agent shown in FIG. 4 includes several rules, one of which is shown in FIG. 5. A rule contains an "IF" field (predicate) and a "THEN" field (action). Both these parts of the rule contain links to other objects which are shown as embedded templates. The IF part of the rule is a description, a special kind of template that describes a set of instances in terms of the values of their fields. The THEN part of the rule is an Action object.

To construct the IF part of a rule, a user (a) clicks on the IF field with the middle mouse button, (b) selects "Descriptions" from the menu presented, and then (c) selects an object type from the tree of object types presented. This causes a description of the appropriate type to be inserted in the rule as an embedded template, and the user can then fill in the fields in this description to specify the values that must appear in particular fields for an object to satisfy the rule. As in the Information Lens, more complex specifications for a field can be constructed by combining strings with and, or, not, and parentheses (i.e., arbitrary Boolean combinations are possible within a field). If specifications appear in more than one field, then all specifications must be satisfied at once for the rule to succeed (i.e., specifications in different fields are implicitly and-ed). As in the other template-based editors in Object Lens, pop-up menus listing likely alternatives for a field are available in editing descriptions.

To specify the THEN part of a rule, a user simply clicks on the THEN field and selects an action from the menu of alternatives presented. These actions are applied to the "current object" (the object matched by the IF part of the rule) in the context of the "current folder" (the folder specified in the "Apply to" field of the agent). In some cases (such as the "Move" action shown here), the user also needs to fill in some fields in the embedded template for the action (e.g., the field specifying where the object is to be moved). The actions currently implemented in rules include the following: "copy" (add the current object to a different folder without removing it from the current folder), "move" (add the current object to a different folder and delete is from the current folder), "delete" (remove the object from the current folder), and "add keyword" (add the specified keyword to the Keywords field of the object). In addition, rules can invoke object-specific actions, including the actions that apply to all objects such as "hardcopy" and "save." Our ongoing work with action objects includes unifying those used in rules with those invoked by the user directly.

The rules are applied in the order in which they appear in the agent's rule folder. Users can create extended reasoning chains by having some rules set characteristics of objects (using the Add Keyword action) which other rules test (by checking the Keyword field).

Embedded descriptions. With the capabilities we have described so far, all rules must depend only on information contained in the objects to which they are being applied. For instance, a rule about a message can depend only on information contained in the message itself. It is often desirable, however, to be able to specify rules that also depend on other information contained elsewhere in the knowledge base. For instance, in the Information Lens system, if a user wanted to specify a rule that applied to all messages from vice presidents, the rule would have to include in the From field, the names of all the vice presidents.

In Object Lens, it is possible to draw upon other information by having descriptions embedded within other descriptions. For instance, the rule shown in FIG. 6 will be satisfied if the message is from any person with a job title that includes "vice president." To apply this rule, the system checks to see whether the string in the From field of the message is the same as the Name of any Person object in the knowledge base that satisfies the description.

Navigating Through the System

The starting point for navigation through the Object Lens system is the Object Lens Icon, a window that shows whether the user has new mail waiting and includes a menu item to Show Basics (show the basic folders included in the system). The system folders accessible through the Show Basics action include: (1) a folder containing all the other folders in the system, (2) a folder containing all the Templates defined in the system (FIG. 3), (3) a folder containing all the agents defined in the system, (4) a folder for each object type containing all the instances of that type in the system, and (5) the New Mail folder, into which new mail retrieved from the mail server is automatically inserted. In addition, we have designed but not fully implemented two other folders: (6) Everything, a virtual folder containing all objects in the system, and (7) Orphans, a virtual folder containing all objects to which no links exist.

These basic folders provide users with convenient starting points for locating any object in the system. In relatively small systems, users can browse through these folders directly. In larger systems, we expect that users will let their agents search through the system folders to find objects that meet certain criteria. It is also possible for (a) individual users to create their own customized "directory" folders that contain the folders and other objects they most often use, and (b) applications developers to create folders containing the objects used in their application.

Saving and Sharing Knowledge

One of the important efforts we are pursuing in the Object Lens system involves different ways for people to save and share the kinds of knowledge described above. For instance, we are currently experimenting with linking Object Lens to a remote database server that contains large shared relational databases. This work is still at an early stage, but it is clear that the usefulness of Object Lens will be significantly enhanced if it includes access to shared databases. In the current version of Object Lens, we have preliminary solutions to the problems of saving and sharing knowledge that meet some, but not all, of the needs people will have in this area.

Saving knowledge. Users can save an object (or collection of objects in a folder) at any time by performing the Save action on the object (or the folder). This action uses the file package commands from the underlying Loops and Lisp systems to store the objects in permanent files in a form that can be reloaded at any time. There is also a "Save" action on the main Object Lens icon that saves all the instances in the workstation.

The potential disadvantages of this approach to saving knowledge are that (1) it requires explicit user actions to save objects to permanent storage and (2) it requires all knowledge used by the system to be loaded onto the local workstation. Sharing remote databases will, of course, help solve these problems, but we expect that systems like Object Lens can be of value even without shared databases. For example, many users are already accustomed to explicitly saving their work in applications such as word processing, and even this task can be simplified by creating agents to run periodically (e.g., every night) and do automatic backups of selected objects.

Sharing knowledge by sending messages. There are two ways users of Object Lens can share objects with each other: (1) by sending messages, and (2) by transferring files. When an Object Lens user sends a message, the message object is converted into text and sent via the existing mail system. Any connected electronic mail users can receive and read this textual message. When an Object Lens user receives the message, it is added as a new object in the receiver's knowledge base.

When a user sends a message containing an embedded object that is expanded (as in FIG. 2), the embedded object is converted into (indented) text in the message in a form that (a) can be easily read by any receivers who are not using Object Lens, and (b) is reconverted into another embedded object when it is received by Object Lens users. When a user sends a message containing embedded objects that are not expanded (e.g., that are shown only as link icons), the names of the objects are included in the message in place of the link icons, but these names are not resolved back into icons at the receiver's end. However, certain types of messages, called "envelopes," arrange to transfer copies of unexpanded, embedded objects to the receiver's work station by automatically saving and loading them.

One intriguing research direction here involves how to communicate embedded objects in such a way that they can be resolved into pre-existing objects at the receiver's end. For example, if the sender's message contains a link to a person object, it would be nice for the receiver's system to be able to automatically resolve this link into the receiver's object representing the same person.

Sharing knowledge by transferring files. The second way for users to share objects is by transferring files. As described above, it is easy for users to store on a file server the current state of a set of objects. Other users can then load these files to create (or update) the objects in their own workstations. Saving and loading these files can often be done automatically, such as by using an "envelope" object. For example, we expect that a common way for users to keep current versions of shared information such as names, addresses, and job titles of people in their organization will be to have someone maintain the official version of this information and periodically distribute updates to other users in the organization. Distributing these updates could be done in several ways: (1) the "maintainer" could have automatic agents that periodically store the current versions on a file server and the other users could have automatic agents that periodically load the most recent versions, or (2) the maintainer could explicitly send out messages announcing the availability of files containing updated objects and the other users could have agents that automatically load the files announced in such messages (e.g., a rule might load all files specified in "Official file update" messages from the official maintainer).

One potential problem with this approach is that any changes the users have made to their local copies of objects (e.g., any notes they had added in the Comments field) will be lost when a new version of the object is loaded. To help solve this problem, we are currently investigating more specialized updating actions for agents to use. With this approach, the official maintainer will be able to distribute update messages that specify changes in particular fields of particular objects. Users can then set up agents that make these updates automatically under most conditions, but under certain conditions the user might be notified before the update is made (e.g., if the field about to be modified has previously been changed by the user). In some cases, the user might want to have the change made automatically but also be notified (e.g., if someone in the user's group is changing phone numbers).

Object Lens as a Network User Operating Environment.

User Operating Environments have been around for many years. A traditional model of a user operating environment is shown in FIG. 21A. An application is launched from within an operating environment, which runs on top of the Operating System, which controls the hardware. The Macintosh Finder(TM) is an example of this model. With the advent of networks, this simple model evolved to incorporate the new hardware, as shown in FIG. 21B.

Users are able to take partial advantage of the network by working through new applications such as terminal emulators, electronic mail, and distributed database systems. However, the full benefits achievable through the network await a new model, as shown in FIG. 21C. By creating a common, connected user environment, Object Lens permits users to share information and coordinate activities more fully than with prior art systems.

3. OTHER APPLICATIONS

In this section, we will give more examples of how the above features can be combined to create a variety of cooperative work applications.

Task Tracking

One frequently mentioned capability for cooperative work applications is the ability to keep track of the tasks people are supposed to do (e.g., Winograd & Flores, 1986; Sluizer & Cashman, 1984). For instance, such systems can help answer questions like: What tasks have other people requested me to do? Are any of these tasks overdue? What tasks have I requested other people to do for me?

It is a straightforward matter to support capabilities like this in Object Lens. For instance, the system already includes message types for action requests and commitments. Even in the Information Lens, it was possible to automatically sort these messages into folders according to who is to perform the task, which project it involves, and so forth. In the Information Lens, however, the summary display of a folder's contents shows only the standard message header fields: From, Date, and Subject. To see more about the tasks, individual messages have to be displayed, one at a time. In Object Lens, the messages within a folder can easily be summarized by displaying whatever fields the user chooses. For example, FIG. 8 shows a table display of action request messages that includes the action deadline.

Intelligent Message Sorting: Engineering Chance Notices

An intriguing example of a cooperative work problem involves disseminating information about changes in product specifications (often called "engineering change notices") to the appropriate people in the organization. It was already possible in the Information Lens to sort engineering change notices according to the contents of fields such as Part Affected, Type of Change, and Severity. In Object Lens, it is possible to use additional knowledge to do even more intelligent sorting. For instance, FIG. 9 shows a rule that uses a doubly embedded description to select all change notices that involve parts for which anyone reporting to a particular manager is responsible.

Database Retrieval

There are clearly many cases in both individual and cooperative work when it is useful to be able to automatically retrieve from a database objects that satisfy certain conditions. Object Lens provides a simple way to perform database queries: Users can simply create agents that scan the objects in one folder and insert links to selected objects into another folder. The rules in the agents specify the criteria for selecting objects.

For instance, suppose you wanted to find all the technical staff members who were assigned to both the project code-named "Dragon" and the one code-named "Lancelot." FIG. 10 shows a rule that would retrieve all such people. Instead of listing all the technical job titles by name ("software engineer," "systems programmer," etc.), the rule includes an embedded description to determine whether a particular job title is on the technical, as opposed to the managerial or administrative, career ladder.

In addition to this general interface for database retrieval, we have also implemented a specialized feature in Object Lens for determining the recipients of messages. With this feature, descriptions (like that shown in the IF field of FIG. 10) can be embedded in the TO and CC fields of a message. Then, when the message is sent, these descriptions are automatically applied to all the Person objects in the local knowledge base and the resulting people are inserted in the TO and CC fields. This feature allows senders to create distribution lists that are dynamically computed at message-sending time based on the current information about people in their data base (see Zloof, 1981 for a similar capability).

Hypertext

As noted above, it is a straightforward matter to use many of the features of a hypertext system in Object Lens (e.g., Halasz, Moran, & Trigg, 1987; Garrett, Smith, & Meyrowitz, 1986; Delisle & Schwartz, 1986). For instance, our system currently contains an object type called Text that displays only two fields: Name and Text. The Text field of a Text object can contain links to as many other objects as desired. For example, FIG. 11 shows a sample Text object that contains links to people and bibliographic citations as well as to another Text object.

In addition to the usual benefits of hypertext systems, Object Lens derives additional benefits from its integration of hypertext with other database, messaging, and computational capabilities. For instance, in order to insert a link to another node in a hypertext system, a user must first find the node to which the link will be made. In Object Lens, the database retrieval capabilities described above can be used to automatically find objects (such as people or bibliographic citations) that satisfy certain criteria. Then links to these objects can be inserted into the text. One desirable feature found in some hypertext systems that is not yet included in Object Lens is the ability to show and follow the incoming links to an object. We would like to implement this capability as another action available on all objects.

It is interesting to observe that Object Lens appears to have some functionality in at least four of the seven areas that Halasz (1987) listed as being needed in the next generation of hypermedia systems (search and query, computational engines, collaborative work, and tailorability).

An Example

Representing the Structure of Arguments

To illustrate the benefits of adding semistructured objects to hypertext, we will describe an increasingly powerful series of sample applications in Objects Lens, each adding a class of benefits not present in the ones before. All these sample applications are designed to help people represent and manipulate the structure of arguments. For instance, a team of programmers designing a new computer system might use such applications to represent alternative design choices, the arguments for and against each one, and the rationale for the choices finally made. The application could thus help make the decision rationale for later reference, such as when the program is being modified.

A number of systems have been built or proposed to provide this general kind of functionality Conklin & Begeman, 1988; Stefik, Foster, Bobrow, Kahn, Lanning, & Suchman, 1987; Lowe, 1986; Smolensky et al, 1987; Lee, 1989). Our examples will be modeled primarily on the gIBIS system, since it is a recent and appealingly simple example of this class of systems.

Our primary purpose here is not to extend the state of the art in argumentation systems, but to show how easily many of these capabilities can be implemented in a general framework like ours that include semistructured objects. In all but the very last examples, these capabilities can be implemented in Object Lens by knowledgeable end users in a matter of minutes or hours, with no special programming required. For example, our first implementation of most of these capabilities required less than two hours and used only the features of our system that are exposed to non-programming users.

Step 1

"Simple" Hypertext

FIG. 12 shows an example of how a "simple" hypertext system might support our argumentation application. This example is from the Object Lens system, but it uses only unstructured text nodes and links between nodes. Note that it is entirely up to you, the user of this system, to structure the information in the node and to put in the proper links. (In this example, and throughout the discussion in connection with FIGS. 12-20, we will use "you" to refer to users when that simplifies the explanation of system features.)

Step 2

Creating Individual Nodes

The next step in our series is to create semistructured object types that are appropriate for our application. For instance, the gIBIS system has three types of nodes: Issues, Positions (about an issue), and Arguments (which support or object to positions). FIG. 13 shows an argument node that supports one position and objects to another one. The relationships between this node and the positions to which it refers are represented by links in the appropriate fields ("Supports" and "Objects to", respectively). When users click on a field name, a menu pops up containing suggested alternatives for that field. For instance, when you click on the Keyword field a list of suggested keywords appears and if you select one, it is automatically inserted in the field.

To define the new node types (Issue, Position, and Argument), you specialize existing object types. For instance, you can create the new node types as specializations of Thing. To do this, you select the "Create Subtype" action on the type definition for Thing and then use the "Add Field" action on the new type to add the new fields (e.g., Supports, Objects to, and Author) that are present in this type, but not in Thing. In this case, the user also used the "Edit Field Names" action to rename the "Name" field (that was inherited from Thing) to be "Subject". To change the alternatives that are suggested by the pop-up alternatives menu, you can use the "Edit Alternatives" action. This action allows you to either directly edit a list of alternatives for a field, or to specify a folder whose contents at run-time will be displayed as the alternatives.

At this point, the system helps you create nodes with appropriate fields and alternatives, but it doesn't help you view the structure of the argument as a whole.

Step 3

Summarizing the Contents and Relationships in Groups of Objects

The next step is to add ways of displaying the overall structure of the argument using customizable folders. The original gIBIS system has a graphical browser that shows a tree format display of the nodes in the argument, and FIG. 14(a) shows a similar display from Object Lens.

To create such a display in Object Lens, you can simply put all the nodes you wish to have displayed into a folder, select the "tree" display format for the folder, and select which links you want to have shown in the tree. For instance, in FIG. 14(a), the user chose to display links from the Supports, Objects to, and Responds to fields. As the figure shows, links from each of these different fields is shown with a different kind of line (e.g., solid, dashed, etc.). (Note: It is more convenient to specify the display format for this folder if you first create an abstract type, say "gIBIS Node," whose specializations are Issue, Position, and Argument, and which contains all the fields contained in any of its children.)

It is also easy to display the same nodes in a table format display, selecting whichever fields you wish to be shown in the table (see FIG. 14(b). FIG. 15 shows an additional feature of folders: the ability to select objects from another folder to be inserted into this one. For instance, the selection rule in FIG. 15 selects all the counterarguments to a specific position.

At this point in the example, the system has most of the basic user interface functionality of the original gIBIS system. (Unlike the original gIBIS system, however, we have not implemented connections to a remote database server, nor have we hardened the system to the point where it can be used reliably by a large group of people.)

Step 4

Automatically Selecting and Manipulating Objects

The last step in our example is to add intelligent agents to help search and modify the network of nodes. For instance, FIG. 16 shows an agent like one you might use to notify you whenever people add arguments that support positions you have entered. This agent is triggered automatically when new objects are added to the folder containing the discussion of interest. FIG. 17 shows the rule this agent uses to select the arguments that support a specific person's positions. This rule illustrates how embedded descriptions can be used to specify structural queries that depend on the link structure in the network as well as on the contents of individual nodes. FIG. 18 shows another (multiply embedded) rule that selects arguments that object to positions authored by people in a particular group (i.e., people who have a specific supervisor). This rule illustrates how queries can use information from throughout a user's knowledge base (such as knowledge about people and their supervisors).

Specialized Display Formats: Calendar

Most of our work so far has used two display formats for folders: tables and trees. These formats are applicable to a very wide range of object types and relationships. It is also possible, however, to create more specialized folder display formats for special kinds of objects. For instance, FIG. 19 shows a specialized display format called "Calendar." The calendar format is used to display objects of type "Event". All events (including specializations such as Meeting and Seminar), contain fields called "Date," "Start time," and "End time." When events are displayed in a calendar, these fields are used to locate the event in the proper day of a month (FIG. 19(a)) and at the proper item within a day (FIG. 19(b)). Users can select a month display like that in FIG. 19(a); they can click on one day within the month display to bring up a day folder like that in FIG. 19(b); and if they display an event within the day folder, the event itself is shown as in FIG. 19(c).

Automatic Agents for Searching and Manipulating Networks

In addition to summarizing the contents of semistructured objects, the system can use their structure to perform even more powerful automatic actions such as searching and restructuring. The Object Lens system uses rule-based agents to perform these automatic actions. For example, FIG. 20 shows an agent that maintains a folder of "Overdue Tasks." Every night at midnight, this agent is automatically triggered and searches the "*All Tasks" folder, a system-maintained folder that contains all task objects in the local workstation. When the agent finds tasks whose due date has passed, it moves them into the Overdue Tasks folder.

4. SYSTEM ARCHITECTURE

What is "Object-Oriented"

Object Lens is an objected oriented, event-driven program. An event-driven program is one where what goes on in the program is driven by user acts (keyboard, mouse, etc.) or other external acts (e.g. the arrival of mail). An object-oriented program has the following characteristics:

Data is encapsulated within objects which insulate the user from the details of how that data is accessed.

Each object is a member of a class. Classes are organized into an entangled hierarchy. Thus, each class has a super-class or parent from which it inherits certain characteristics and methods which instance-objects of the class use for retrieving, changing, or presenting the data in the object.

In an object-oriented program, the program "flow" consists of objects sending messages to each other. Each message, in essence, asks the object to invoke a method designed to respond to that message. Thus an object sent a "Tell-me-the-value-of-your-name" message would invoke a method which is part of the object's class (and which may have been inherited from the object's class's parent) to determine the object's name and deliver it to the message's sender. Similarly, an object sent a "Display-yourself"message would invoke a method belonging to or inherited by the object's class which can present the data in any instance of the object's class.

Overall Architecture

The Object Lens system can be thought of as having a number of modules which perform specific tasks. These modules and their relationships are illustrated in FIG. 16.

Object Manager

The heart of Object Lens is the Object Manager. Currently built on top of LOOPS, the Object Manager is responsible for keeping track of all classes and class-instances and their links to each other. It also keeps track of the current state of each object and helps the objects handle messages which they receive by providing support functions for their methods. The Object Manager provides the Forms Manager with the information it needs to present a form. The Object Manger also handles saving and loading objects from permanent storage in the database. In the future, the Object Manager will work with a shared database to do object locking and version control.

Forms Manager

When an object receives a "Present Yourself Using Form X" message, it executes the appropriate presentation method by calling the Forms Manager. The Forms Manager also handles user-events addressed to displayed forms and, where necessary, translates them into messages addressed to the appropriate object(s). Examples of messages include:

The forms manipulated by the Forms Manager are built on top of a set of complex human-interface toolkit objects, which are, in turn, built on top of a window system. While the current implementation of Object Lens uses the toolkit procedures available on the Xerox Interlisp Workstation, the concepts are intended to be portable to other environments, such as X-Windows.

Agent Manager

The Agent Manager knows about each agent's automatic triggers. It includes processes that watch for time-based triggers and receives messages from the Object Manager about New Links and Object Updates. It also receives messages from the Object Manager about agents which have been manually triggered.

Triggering an agent causes the agent to perform its "trigger" method. It does this by determining to which folder it applies, picking out each object in that folder sequentially and telling the rules in its rules folder to apply themselves to the object. Rules use a recursive pattern matcher to determine if an object matches the description contained in the rule's left hand side, and if so sends an "Execute" message to the actions on the rule's right hand side.

The recursive pattern matcher compares the description on the left hand side of a rule to a particular object. If the type is the same (or the object is a sub-class of the description's type), the pattern matcher performs a case-insensitive, substring match against the contents of the description's field. If a description-field contains an embedded description and the object being matched has an object within the same field, the pattern matcher calls itself to see if the embedded description matches the linked object.

Mail Manager

The mail manager handles incoming and outgoing messages. It translates incoming E-Mail into Object Lens objects and accepts the "message stream" produced when an object executes its "send" method. The current Mail Manger is built on top of the Xerox Lafite system, but could be built on top of other mail systems.

In the future, the Mail Manager will also coordinate the shipping of objects between disjoint Object Lens installations.

Database

Currently, permanent storage of objects is only through ASCII files which must be saved and loaded.

Lens Object Classes

In Object Lens, the internal object-class structure is designed to reflect the objects with which the user may interact. The pre-defined object classes in the system are shown in FIG. 13.

All Lens Object Classes have internal attributes which are used when creating instances of the Class. In addition, Lens Object Classes have methods which are used to manipulate the data in the class's instances. In particular, Lens Object Classes have presentation methods called Forms. Each Class may have one or more Forms; the Form that is displayed depends on the context in which the Class-instance is being shown and on the which Form is considered to be the default for that Class. All internal attributes and methods of a parent-Class are inherited by its child-classes.

We will go through each object class and describe its attributes and presentation methods (Forms).

Form

All objects in object lens are presented through forms. While most objects, including classes, have only one form through which they may be presented, instances of some types can be presented through multiple forms. Forms consist of two parts: the control area and the data areas; forms differ primarily in their data areas.

The control area is at the top of the form. The upper part of the control area is where the Form Manager places notices and user prompts and is also used to move and re-size the form's window.

The lower part of the control area has "buttons" for commands which affect either the form itself or the object(s) shown in the data area(s) of the form. It also has a button called "*Other*" which pops-up a menu of less frequently used commands. Forms maintain ordered lists of commands. The one of these lists which is "active" is used to determine what commands will be available: The Form Manager puts as many of the items on the active list onto the button region as will fit, given the window size. The rest are put into the *Other* menu.

When an object receives a request that it "show" itself, it first determines what form it should be presented through. (Note: the generalized method of using forms as described here is not yet fully implemented in our prototype system.) If an appropriate form was not specified in the message, the object picks its default form. The object then asks the object manager to create or re-use an existing form-instance of the appropriate class. The object then sends a *Show message with all the appropriate object-data to the form instance. From then until an Update Object message is sent to the form, the user interacts with the data in the form and not with the original object. The form determines which list is active by looking at the value of one or more fields in the object to be displayed. This allows the form manager to present different command options depending on the value of status attributes of the object being displayed. A form may have zero, one, or more data areas. At the top of each data area is a label; below the label is where the data about the instance is presented. Form, like other classes, may be sub-classed; many different sub-classes of Form are described below. Currently, the user cannot create Form sub-classes. However, this will change in the future.

System

System (or "Executive") is a class of which there is, currently, only one instance-object. The Form for the System instance is a Control Panel, which has no data area. The form's command list has six items; the *Other* button is re-labeled "Show Basics." The default window-size allows only the first of these commands to be shown on the Button Bar:

Thing

Thing is a class which has a meta-class called Type. Looked at another way, Thing and its sub-classes are both classes and types (instances of Type). While this may seem confusing, it is both important to understand and very powerful. Because users can create instances of Classes, and because Type is a class, users can create types. Since Thing and its sub-classes are types, users can "sub-class" Thing. This is done through a "types folder," which allows users to create types which are sub-classes of Thing. As shown in FIG. 13, Thing's predefined sub-classes include Message, Rule, Folder, and Agent, which are discussed in detail below. However, it is part of the power of Object Lens to permit the definition by the user of a type of which an instance may be a conventional data record.

Naturally, user-made types inherit their parent type's Forms and attributes. Users can create any number of instances of any type. Instances of each of these classes behaves in the same way as instances of Thing, except that the number, nature, and name of each of the fields may be different. Types can have any number of fields and use, by default, the "Template" sub-class of Form.

As shown in FIG. 15, Thing starts out with two shown fields: Name and Comments. The Keywords field is not displayed in this form.

A template is a form in which the data-area consists of a series of field-labels followed by slots into which data may be entered. Into these slots the user may enter both free text and icons which serve as a hypertext-link to another object.

A slot can also contain an embedded object which appears as the data-area of a template for another object. Such an embedded template is a hyper-textual link to another object; the other object can be edited right there within the template for the linking object. A link to another object can be displayed either as an embedded template or as an icon, depending upon the Is-It-Expanded attribute of the field.

Mousing left on the field labels pops-up a menu of alternatives for the field's value. Mousing middle on the field label gives the user access to the explanation, default value, descriptions, and alternatives for that field.

A template's command list contains the following commands:

Message

Message is a predefined type of Thing which includes two additional shown fields, To: and From:, and renames the first field, Name, which is inherited from Thing, "Subject." The text of the message goes in the second field inherited from Thing.

The message template has two command lists. The first is used when the Creator field of the message is empty (i.e. before the message has been sent); the second, when the field has something in it (i.e. after the message has been received). The first command-list is as follows:

The second command list is as follows:

Naturally, one may create sub-classes of Message. These inherit the additional characteristics of Message.

Rule

A rule has three fields which are shown: Name, If, and Then. The If field takes descriptions as values. The Then field takes actions as values. These actions are pre-wired into the Then field as alternatives.

A Selection Rule is a sub-class of Rule. Instances of Selection Rule can belong to a folder. Other than that, selection rules do not differ from regular rules.

Folder

Folders are containers and are one of the most powerful features of Object Lens. Like Thing, users can create instances of Folder. The most important attribute of folders is that they contain a field which contains a list of links to other objects. Folders also have a type of object that they prefer to contain; the user is asked to identify this type when a new folder is created. Finally, folders can also have a selection rule which can be used as a kind of "agent on special assignment" to collect objects to put into the folder.

Folders can be displayed using instances of a variety of sub-classes of Form. While only two sub-classes have been implemented so far (Tables and Graphs), others, including Calendars, Desktops, and Outlines, will be added in the future.

A table's data area shows a row for each item contained in the folder. The rows are divided into several columns in which fields from the objects are shown. What fields are "shown" are determined by a menu-selection reached through the "Choose Fields to Show" command. Rows may be selected using the left and right mouse button. The user can also use the middle mouse button on a row to bring up a menu with the following context specific options along with any object specific actions:

A graph's data area shows icons for each object contained in the folder. The icons are arranged as the nodes in a graph; the lines of the graph represent the object-field(s) the user tells the graph to "show." What fields are "shown" are determined by a menu-selection reached through the "Choose Fields to Show" command. If more than one object-field is to be shown, the graph displays the lines using different line patterns. The geometry of the graph is determined by the lines and by order in which the objects were entered into the folder.

Icons may be selected using the right mouse button. The user can also use the middle mouse button on a row to bring up a menu with the following context specific actions along with any object specification:

Types Folder or "View" Folder

Instances of Types Folder, such as "All Templates," use instances of a sub-class of Graph called a Types Graph. A types graph differs from other graphs in the behavior of objects in its data-area: the middle button used on an icon representing a type brings up a menu with the following options:

Using the left button on an icon for a type causes an instance of that type to be created and to be presented using the type's default presentation form.

Agents

Agent is a pre-defined class of objects. Agents contain a field for:

the folder to which the agent is to be applied,

the list of things which can automatically cause the Agent Manager to trigger the agent (such as new links to the folder,the updating of an object contained in the folder, the passing or an hour, or the passing of a day), and

a folder containing a set of rules. The commands list of an agent form contains:

Application Document

This type of Thing has not yet been implemented. However,it offers one model of how documents from other applications could be encapsulated in Object Lens. An Application Document object would have a template-form. However, the form would have an additional command in the control area which would allow the user to select another form to display the document; that form/method would be the document's application. The object itself would only contain a reference to the file-path of the application document in a hidden field. Choosing the other method would launch the application with the document in a separate window.

Type

An instance of class Type is an object which serves as the class definition for another object. In particular, the class Thing and all of its sub-classes are instances of the class Type. A type contains a list of fields. Each field in a thing has several attributes:

A type is displayed through a set of forms which are sub-classes of Template and through pop-up menus, each of which allows the user to edit different attributes of the fields. These forms are:

The user sets whether a field is to be Shown or Expanded through pop-up menus which are available through any of the above forms. The user sets descriptions for a field through a series of pop-up menus.

Description

The Pattern Matcher uses descriptions to identify objects which match the descriptions criteria. A description for a thing is identical in appearance to the template-form for that thing, except the font of the text in the description is in italics. A user can fill-in a description.

Variables

A variable is a special container-object which is useful in descriptions and actions. (Note: this use of variables is not yet implemented in the prototype.) Just like any other object, it has a name and appears as an icon when inserted into another object which is being shown through a template-form. A user can also show a variable through a template-form which is a simple window showing the variable's name and current value. A user may insert a new variable into a description at the selection point by using the Add Link message. When a variable is encountered in a field description during a matching, it is replaced with its value unless it is the first argument of a Set Variable expression. A variable's value may be set in one of two ways. The user may edit the variable through it's form, changing the variable's value and/or name. The user may also include a Set Variable expression in a description. Since Matching is done from top to bottom, the value of a variable at any point in a description (or rule) is equal to the value set by the closest Set Variable expression before or above the current point or, if there are no SetVariable expressions above or before the point, the variable's value external to the description.

Actions

Action are objects which specify processes to be applied to other objects. Actions are associated with particular types of objects. Some actions (such as "Show" and "move") are predefined for specified object types. In future versions of the system, we expect to let users (and application programmers) define new actions for themselves. For instance, it will be possible to define new actions for a specific object type (and its specializations) by writing procedures in a general purpose programming language (such as C or Lisp). We also plan to have simplified programing languages that "non-programmers" can use to define new actions that combining actions already defined.

One possible such language which we have partially specified but not yet implemented is called Object Action Language (OAL). An OAL expression consists of a message, a target object, and the arguments which the message takes. Any user-method for any object may be used as part of an OAL expression. If an expression returns a result (e.g. Create Object returns a pointer to that object), that result is placed in a special variable called "The Result." An OAL expression may also call other actions by sending them an execute message. The THEN field of a rule can also contain OAL expressions which are evaluated when the rule fires. Actions are objects which contain procedures which are executed when they receive a Execute message. The message may be sent to the Action either as a result of a rule's firing or by the Forms Manager as a result of a command's being executed. Procedures are written in terms of a limited set of primitives, such as:

Add Link from folder () to object ()

Remove link from folder () to object ()

Set field () of object () to be equal to string/object/result-of-function ( )

Put the value of field () into variable ()

Create an instance of class () {and store its name in }

Delete object () from virtual memory

Pass object () to the Mail Manager for shipping

Save object () to permanent storage

Retrieve object() from permanent storage

Expunge object () from permanent storage

Hardcopy object () {using form () }

Sort folder () on field ()

Trigger agent/rule ()

Execute action ()

The ()'s can contain either constants or variable names. The current implementation hard-codes all actions and commands. Thus, rules may only execute the actions:

Move this thing to folder ()

Copy this thing to folder ()

Remove this thing from this folder

Hardcopy this thing

Resend this thing

Add keywords () to this thing where each action is performed on the "current object" (selected by matching the IF part of the rule) and in the context of the "current folder" (the one to which the agent applies).

Actions do not currently have presentation forms but will, in the future, have editor-forms geared towards procedure development.

Commands

It is useful to distinguish between actions and commands. A command is an instruction given through a form. The instruction may be invoked through the button-bar or through a pop-up menu. Commands may apply to a window, a form, or to one or more objects. Window commands, such as Close, Move, and Shape, are typically accessed through a pop-up menu, are handled directly by the Window System. Form commands, such as Change Display Format, or Edit Fields to Show, or Select are typically accessed through the button bar and are handled by the Forms Manager. Object commands, such as Show, Add Link, Delete, Send, Answer, and Forward can be accessed either through the button bar (when the object of the action has been selected in advance) or through pop-up menus (when the menu is gotten by clicking-middle on the object of the action). Object actions are also handled by the Forms Manager, which translates them into action requests to the Action Exec.

The Delete Selection command is a useful example. After the user has issued a series of commands to the form in order to select a set of objects contained in a folder and being shown by the form, the user goes to the button-bar and chooses Delete Selection. The Forms Manager uses its list of selected objects to translate this command into a series of Delete actions which changes the contents of the folder to remove the selected items. Like actions, commands are also currently hard-coded. Also like actions and forms, in the future, users will be able to write commands and incorporate them into forms.

Detailed Technical Documentation

This section describes the detailed System Specification for the Object Lens system as currently implemented on top of LOOPS in Interlisp D on Xerox Lisp Workstations. It consists of approximately 375 Kb of unique source code, independent of the system code (Interlisp, LOOPS, Lafite, TEdit, Grapher, TableBrowser) upon which Object Lens is based. The unique executable code takes up about 250 Kb of virtual memory; the starting knowledge base takes up about another 15 Kb. Because Object Lens is implemented using object oriented programming, the system architecture must be understood in terms of the classes of objects which have been established.

Organization of the LOOPS Classes in Object Lens

The structure of the Object Lens program in accordance with a preferred embodiment of the invention is shown in FIG. 17. The top level, non-mixin class is Lens Object. It contains the most general information about all Lens Objects. The classes known to the non-programmer "user classes"! start from Thing. All user classes have Type (Template) as their meta-class. A user class contains a description of each of the fields to be used by an instance of that class. The attributes of a field include its name, its customized name, its default value, the alternatives to be shown in its alternatives menu, and the descriptions which may be inserted into it. Fields are inherited from parent user-classes but may be modified locally.

A class also contains three kinds of methods: User-Action Methods (e.g. Show, Close) are accessible to the user through button-bar or pop-up menu items and are typically attached to the class. Presentation methods (e.g. show an object through a template-form) are typically attached to one of the mixins. Other System Methods, such as Prepare Sending, perform system utility functions and are also typically attached to a mixin.

Methods by Class

The methods listed below are either system or presentation methods (which start with a `*`) or user methods which, with the exception of `*Other*`, do not start with a `*`.

Editor Mixin

This mixin contains many presentation methods used by all classes in object lens when an object is presented through a Template-Form. The mixin also contains some methods which are useful for all objects in Object Lens which have nothing to do with presentation.

Methods for Presenting a Template-Form

Methods for Manipulating the contents of a Template-form

Methods for Updating an Object from the Contents of a Template-Form

Methods which Only Affect the Saved Object

TableMixin

TableMixin contains methods which are only used by the Folder object. The methods are stored in TableMixin for code management reasons only. TableMixin was not made a sub-class of LensObject because its methods are, in some sense, applicable to systems other than Object Lens.

TreeMixin

TreeMixin contains methods which are only used by the Folder object. The methods are stored in TreeMixin for code management reasons only. TreeMixin was not made a sub-class of LensObject because its methods are, in some sense, applicable to systems other than Object Lens.

Lens Object

BrowserMixin

BrowserMixin contains methods which are only used by the Folder object. The methods are stored in BrowserMixin for code management reasons only. BrowserMixin was made a sub-class of LensObject only because its methods are specific to Object Lens.

Executive

This is the "control panel" which says "Object Lens" on it.

Action

Action's Progeny

Template (Type)

Folder

Certain functionality, like "Add Link," requires a different method depending on whether the folder is being displayed through a table or a graph. This is due to a limitation in LOOPS which prevents the inheritance path from being modified.

Agent

Rule

Single Instance Classes

There are two classes which have a single, pre-defined instance ("system instances"): *Executive and *New Mail. In addition, there are so called "virtual folders" which do not exist at load-time but which are created by the system and filled with all instances of a particular type upon the user's request. Virtual folders include: *All Types, *All Agents, *All Folders, and *All XXXXXs where XXXXX is any user-class of thing.

The Organization of the Files

Each file in the system is listed, followed by a brief summary of its contents.

Additional Requirements

The Object Lens code attached to this application assumes that the following changes have been made to the standard Xerox Lisp environment, containing TEdit and Loops:

(1) The following three file package functions are modified to allow spaces in class names:

(a) MethodFileOut

(b) ObjectFileout

(2) The following interfaces with a mail transfer mechanism are provided in a way that is equivalent to the Lafite mail system:

(a) Definitions of the following:

(i) variables

(ii) records

(iii) functions (required)

(iv) functions (optional)

These functions provide the following functionality:

References

Conklin, J. (1987). Hypertext: A survey and introduction. IEEE Computer, 20(9), 17-41.

Conklin, J. & Begeman, M. L. (Sep. 26-28, 1988). gIBIS: A hypertext tool for exploratory policy discussion. ACM Transactions on Office Information Systems, 6, 303-331.

Delisle, N. & Schwartz, M. (1986). Contexts--a partitioning concept for hypertext. ACM Transactions on Office Information Systems, 5(2), 168-186.

diSessa, A. (1985). A principled design for an integrated computational environment. Human Computer Interaction, 1, 1-47.

diSessa, A. & Abelson, H. (September 1986). Boxer: A reconstructable computational medium. Communications of the ACM, 29, 859-868.

Dittrich, D. & Dayal, U. (Eds.) (1986). Proceedings of the International Workshop on Object-Oriented Database Systems, Asilomar, Calif.

Fikes, R. (Aug. 23, 1988). Integrating hypertext and frame-based domain models. AAAI 1988 Workshop on AI and Hypertext, St. Paul, Minn.

Fikes, R. & Kehler, T. (1985). The role of frame-based representation in reasoning. Communications of the ACM, 28, 904.

Garrett, L. N. Smith, K. E. & Meyrowitz, N. (1986). Intermedia: Issues, strategies, and tactics in the design of a hypermedia document system. Proceedings of the Conference on Computer-Supported Cooperative Work, Austin, Tex. Dec. 3-5, 1986, 163-174.

Goldstein, I. & Bobrow, D. (1987). A layered approach to software design. In Interactive Programming Environments, D. Barstow, H. Shrobe, and E. Sandewell, Eds. McGraw-Hill, New York, pp. 387-413.

Halasz, F. G. (July 1988). Reflections on notecards: Seven Issues for the next generation of hypermedia systems. Communications of the ACM, 3,836-852.

Halasz, F. G. (1987). Reflections on NoteCards: Seven issues for the next generation of Hypermedia. Communications of the ACM, 31 (7), 836-855.

Halasz F. G. Moran, T. P., & Trigg, R. H. (1987). NoteCards in a nutshell. Proceedings of the 1987 ACM Conference of Human Factors in Computer Systems, (CHI+GI '87), Toronto, Ontario, Apr. 5-9, 45-52.

Harp, B. (August 23, 1988). Position paper: Facilitating intelligent handling by imposing some structure on notes. AAAI 1988 Workshop on AI and Hypertext, St. Paul, Minn.

Lai, K. Y., Malone, T. W. & Yu, K. C. (October 1988). Object Lens: A `spreadsheet` for cooperative work. ACM Transactions on Office Information Systems, 6, 332-353.

Lee, J. (1989). DRL (Decision Representation Language) and its support environment. Massachusetts Institute of Technology, Artificial Intelligence Laboratory, Working Paper #325.

Lee, J. & Malone, T. W. (1988a). How can groups communicate when they use different languages? Translating between partially shared type hierarchies. Proceedings of the ACM Conference on Office Information Systems, Palo Alto, Calif., Mar. 23-25, 1988.

Lee, J. & Malone, T. W. (1988b). Partially Shared Views: A scheme for communicating among groups that use different type hierarchies. Sloan School of Management Working Paper, Massachusetts Institute of Technology, Cambridge, Mass. September, 1988.

Lowe, D. (February 1984). Co-operative structuring of information: The representation of reasoning and debate. International Journal of Man-Machine Studies, 23(1), 112-125.

Malone, T. W., Grant, K. R. Turbak, F. A., Brobst, S. A. & Cohen, M. D. (1987). Intelligent information sharing systems. Communication of the ACM, 30, 390-402.

Shriver, B. & Wegner, P. (1987). Research in object-oriented programming, Cambridge, Mass.: MIT Press.

Smolensky, P., et. al. (1987). Constraint-based hypertext for argumentation. Department of Computer Science and Linguistics, University of Colorado Working Paper #CU-CS-358-87.

Stefik, M. & Bobrow, D. (1986). Object-oriented programming. Themes and variations. AI Magazine, 6(4), 40-62.

Stefik, M., Foster, G., Bobrow, D. G., Khan, K., Lanning, S. & Suchman, L. (1987). Beyond the chalkboard: Computer support for collaboration and problem solving in meetings. Communications of the ACM, 30, 32-47.

Turbak, F. A. (1986). Grasp; a visible and manipulable model for procedural programs. Unpublished M. S. thesis, Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology.

Claims

What is claimed is:

1. A computer user interface, comprising: means for graphically representing objects that collectively are instances of a plurality of object types, wherein an object has a plurality of fields of which at least one field may contain information of any type; and means for creating and editing object types, and for creating, editing and displaying each instance of an object type using a template; wherein the means for graphically representing objects, and the means for creating and editing each object type and creating, editing and displaying each instance, present each object to and permit action upon and editing of each object by a user in a substantially uniform way, regardless of type, and wherein the means for graphically representing objects includes means for displaying a summary of a collection of objects showing one or more fields selected by a user from each of the objects.

2. A computer user interface according to claim 1, further comprising: means for permitting objects to be defined hierarchically by the user.

3. A computer user interface according to claim 2, wherein the graphical display of the link includes at least one of a symbol for the second object, and a template associated with the second object.

4. A computer user interface according to claim 1, wherein each object type has a set of actions that can be performed upon instances of the object type, and the objects are arranged in a hierarchy of increasingly specialized types, with each object type inheriting fields and actions from any parent thereof.

5. A computer user interface according to claim 4, wherein a rule is one of the object types.

6. A computer user interface according to claim 4, wherein a folder is one of the object types.

7. A computer user interface according to claim 4, wherein a folder is one of the object types and a rule is another of the object types.

8. A computer user interface according to claim 1, wherein a first object may contain, as part of a value of one of the fields, a link to a second object, and wherein the means for graphically representing the first object graphically displays the link.

9. A computer user interface according to claim 1, wherein a folder is one of the object types.

10. A computer user interface according to claim 1, wherein a folder is one of the object types and a rule is another of the object types.

11. The computer user interface of claim 1, wherein the means for creating and editing object types includes means for adding fields to an object type.

12. The computer user interface of claim 1, wherein the summary is a table view.

13. The computer user interface of claim 1, wherein the summary is a graphical view.

14. A computer user interface, comprising: means for graphically representing a collection of objects of more than one type in a selected one of formats, wherein one format is a table and another format is a graphical view wherein the structure of the graphical view shows relationships among objects in the collection, wherein an object has a plurality of fields, wherein at least one field may contain information of any type; means for creating and editing object types and instances of object types using a template; and means for permitting the user to customize the formats utilized by the means for graphically representing the collection of objects, according to one or more selected fields defined for the object types.

15. A computer user interface according to claim 14, wherein each object has a set of fields and a first object may contain, as part of the value of one of its fields a link to a second object.

16. A computer user interface according to claim 15, wherein the graphical view is a node-arc graph.

17. A computer user interface according to claim 14, wherein the graphical view is a node-arc graph.

18. A computer user interface according to claim 14, wherein a folder is one of the types of objects in the collection and the collection is a folder.

19. An interface according to claim 14, wherein the means for graphically representing includes means for permitting any of the formats to be applied to any collection of objects.

20. An interface according to claim 19, wherein the graphical view is a node-arc graph.

21. An interface according to claim 19, wherein the graphical view is a calendar view.

22. An interface according to claim 14, wherein the graphical view is a calendar view.

23. The computer user interface of claim 14, wherein the means for creating and editing object types includes means for adding fields to an object type.

24. A computer user interface comprising: means for easily creating and editing object types with different fields, wherein information in at least one of the fields of each object may be of any type; means for easily creating and editing instances of objects of the defined object types; and means for easily creating different user-tailorable summary views of collections of the instances of objects, wherein a view is defined by one or more selected fields defined for the object types.

25. The computer system of claim 24, further comprising: means for easily defining links between instances of objects; and means for providing traversal of links.

26. The computer system of claim 25, wherein links may be active links in fields of objects; and wherein the means for creating views may create views of objects generated by following active links in the fields used to generate the view.

27. The computer system of claim 26, further comprising: means for easily defining agents for performing functions on instances of objects.

28. The computer system of claim 27, wherein agents are defined as objects having fields including functions to be performed, and wherein the means for creating views allows a view to be created based on fields of agent objects.

29. The computer system of claim 25, further comprising: means for defining agents for performing functions on instances of objects.

30. The computer system of claim 29, wherein agents are defined as objects having fields including functions to be performed, and wherein the means for creating views allows a view to be created based on fields of agent objects.

31. The computer system of claim 24, further comprising: means for defining agents for performing functions on instances of objects.

32. The computer system of claim 31, wherein agents are defined as objects having fields including functions to be performed, and wherein the means for creating views allows a view to be created based on fields of agent objects.

33. The computer user interface of claim 24, wherein the means for easily creating and editing object types includes means for adding fields to an object type.

34. An object-oriented computer user interface system, comprising: a generic semi-structured object type comprising: a collection of fields, wherein at least one field may contain information which may be of any type; a create subclass method for permitting a user to create a new subclass of an object type; a create instance method for permitting a user to create a new instance of an object type; an edit method for permitting a user to edit the object type; and a display method permitting a user to select fields of an object type for displaying a summary of a collection of objects including objects of at least the object type, thereby providing user-tailorable summary views.

35. The object-oriented computer user interface system of claim 34, wherein the generic semi-structured object type further comprises: a field edit method for permitting a user to add a field to the collection of fields of an object type.

Patent Citations (24)

Patent	Date	Inventor
US4613946	1986-09-01	Forman
US4646235	1987-02-01	Hirosawa et al.
US4648062	1987-03-01	Johnson et al.
US4675829	1987-06-01	Clemenson
US4736308	1988-04-01	Heckel
US4764867	1988-08-01	Hess
US4821211	1989-04-01	Torres
US4849878	1989-07-01	Roy
US4879648	1989-11-01	Cochran et al.
US4891766	1990-01-01	Derr et al.
US4918621	1990-04-01	Nado et al.
US4926349	1990-05-01	Thor
US4930071	1990-05-01	Tou et al.
US4974173	1990-11-01	Stefik et al.
US4982344	1991-01-01	Jordan
US5072412	1991-12-01	Henderson Jr., et al.
US5107443	1992-04-01	Smith et al.
US5115501	1992-05-01	Kerr
US5121478	1992-06-01	Rao
US5159669	1992-10-01	Trigg et al.
US5167011	1992-11-01	Priest
US5208748	1993-05-01	Flores et al.
US5216603	1993-06-01	Flores et al.
US5313636	1994-05-01	Noble et al.

Non-Patent Literature (85)

Dittrich, D. & Daval U. (Eds.) (1986). Proceedings of the International Workshop on Objected-Oriented Database Systems, Asilomar, CA (QA 76.9.D3.I586).
Garrett, L.N. Smith, K.E., & Meyrowitz, N. (1986). Intermedia: Issues, strategies, and tactics in the design of a hypermedia document system. Proceedings of the Conferene on Computer-Supported Cooperative Work, Austin, TX, Dec. 3-5, 1986. pp. 163-174.
Halasz, F.G., Moran, T.P. and Trigg, R.H. (1987). NoteCards in a nutshell. Proceedings of the 1987 ACM Conference of Human Factors in Computer Systems, (CHI + GI '87), Toronto, Ontario, Apr. 5-9, pp. 45-52.
Harp, B. (1988, Aug. 23). Position paper: Facilitating intelligent handling by imposing some structure on notes. AAAI 1988 Workshop on AI and Hyertext, St. Paul, Minnesota.
Lee. J. (1989). DRL (Decision Representation Language) and its support environment. Massachusetts Insitute of Technology, Artificial Intelligence Laboratory, Working Paper #325.
Lee. J. and Malone, T.W. (1988a). How can groups communicate when they use different languages? Translating between partically shared type hierarchies. Proceedings of the ACM Conference on Office Information Systems, Palo Alto, CA, Mar. 23-25, 1988.
Shriver, B. and Wegner, P. (1987). Research Directions in object-oriented programming. Cambridge, MA; MIT Press.
Smolensky, P. et al. (1987). Computer-Aided Reasoned Discourse, or, How to Argue with a Computer, Department of Computer Science and Linguistics, University of Colorado Working Paper #CU-CS-358-87.
Malone, T.W., Grant, K.R., Lai, K.Y., Rao, R., and Rosenblitt, D. (in press). The Information Lens: An intelligent system for information sharing and coordination. In M.H. Olson (Ed.), Technological support for work group collaboration, Hillside, N.J.: Erlbaum (DEWEY HD 66,T44 1989).
Sluzier, Suzanne and Cashman, P.M. (1984). XCP: An experimental tool for supporting office procedures. IEEE 1984 Proceedings of the First International Conference on Office Automation, Silver Spring, MD: IEEE Computer Society, pp. 73-80.
Winograd, T. and Flores, F. (1986). Understanding computers and cognition: A new foundation for design, Norwood, NJ: Ablex, (BARKER QA76,W58).
Trigg, R., Suchman, L., and Halasz, F. Supporting Collaboration in NoteCards, Conference on Computer-Supported Cooperative Work, MCC Software Technology Program, Austin, TX (1986).
Goldstein, I.P. and Bobrow, D.G. Layered networks as a took for software development. Proc. 7th Int'l Conf. On Artificial Intelligence, 1981.
Nii, P. The blackboard model of problem solving. The Al Magazine. Spring pp. 38-53.
Skarra, A.H. and Zdonick, S.B. The management of changing types in an object-oriented database. In Proc. OOPSLA, Sep. 1986, pp. 483-495.
Nicole Yankelovich and Norman Meyrowitz, "Reading and Writing The Electronic Book," The Institute of Electrical and Electronics Engineers, Inc., Computer (1986), pp. 15-30.
Crowston, K. et al., "Cognitive Science and Organizational Design A Case Study of Computer Conferencing," in Human-Computer Inter-action, vol. 3, pp. 59-85, 1987-1988.
Ackerman, M., et al., "Intelligent Agents, Object-Oriented Data-bases, and Hypertext," in Proc. AAAI 1988 Workshop on AI and Hypertext: Issues and Directions, Mark Bernstein (ed.), Aug. 23, 1988, pp. 1-3.
Terveen, L. G. et al., "A Collaborative Interface For Editing Large Knowledge Bases," in AAAI National Conference on Artificial Intelligence, Boston, MA: AAAI, 1990, pp. 491-496.
Hollan, J. et al., "An Introduction To Hits: Human Interface Tool Suite," MCC Technical Report No. ACA-HI-406-88. dated Dec. 1988.
Engelbart, D.C. et al., "A Research Center for Augmenting Human Intellect," in Proceedings of the Fall Joint Computing Congress. AFIPS press, Reston, Virginia, 1968, pp. 395-410 Dec. 1968.
Lai, Kum-Yew, "Essays on Object Lens: A Tool for Supporting Information Sharing," S.M. Thesis, Massachusetts Institute of Technology, Dated Aug. 1987, Published Jul. 22, 1988.
Lai, Kum-Yew, "Object Lens: A `Spreadsheet` for Cooperative Work" Sloan School Working Paper SSM #WP 2053-88, Dated Mar. 1988, publication date unknown.
Thomas Malone et al., "What Good Are Semi-Structured Objects?", Sloan School Working Paper WP #3064-89-MS, dated Jun. 1989, publication date unknown.
Ashton-Tate, "Getting Started: dBase III Plus," 1987.
Ashton-Tate, "Programming with dBase III Plus," 1987.
Ashton-Tate, "Using dBase III Plus," 1987.
Ashton-Tate, "Learning dBase III Plus," 1986.
Ashton-Tate, "dBase III Plus: Quick Reference Guide," 1987.
Ashton-Tate, "dBase Bridge, dBase III Plus," 1986.
User Manuals for Ashton-Tate's dBase III Software Product, 1984.
"dBase III: Relational DBM for 16-Bit Computers," Byte: Nucleus: What's New:; p. 50, Jul. 1984.
"DBase ships; dBase for Windows and dBase 5.0 from Borland," DBMS. vol. 7, No. 12, p. 40, ISSN, Nov., 1994.
"DBase 5.0 for Windows: Borland finally delivers the Windows version of the DBMS," Which Computer? vol. 17, No. 8, p. 30, Aug., 1994.
Ricciuit, Mike, "Learning Curve makes dBase 5.0 a dud; Borland says it simple ain't so," InfoWorld, p. 21, Oct. 31, 1994.
"dBase for Windows Ships with Extensive Third-Party Support: Borland Involves Partners in Product's Development, Training, and Quality Assurance," Business Wire, Aug. 9, 1994.
Varney, Sarah E., "Object tools boosts Windows dBase," PC Week, vol. 11, No. 16, p. 6, Apr. 25, 1994.
Gallagher, Sean, "Windows Version of dBase to hit fed desks this summer," Government Computer News, vol. 13, No. 14, p. 1; Jul. 1994.
Newsome, Clare, "Long-awaited Windows dBase hits final delay," PC User, No. 238, p. 12, Jun. 29, 1994.
"Borland Launches dBase for Windows," Newsbytes, Jun. 28, 1994.
"Borland ships dBase for Windows," Business Wire, Aug. 9, 1994.
A. diSessa, "A Principled Design for an Integrated Computational Environment," Human-Computer Interaction, 1985, vol. 1, pp. 1-47.
A. diSessa and H. Abelson, "Boxer: A Reconstructible Computational Medium," Comm. of the ACM, Sep. 1986, vol. 29, No. 9, pp. 859-868.
I. Goldstein and D. Bobrow, "A Layered Approach to Software Design," In Interactive Programming Environments, D. Barstow, H. Shrobe, and E. Sandewell, Eds. McGraw-Hill, New York, pp. 387-413, 88.
I. Greif, "Computer-Supported Cooperative Work: Breakthroughs for User Acceptance," (Panel description), Proceedings of the ACM Conf. on Human Factors in Computing Systems (CHI '88), Washington, D.C., May 16-18, 1988.
Frederich N. Tou et al., "Rabbit: An Intelligent Database Assistant," Proceedings of the National Conference of the American Association of Artificial Intelligence, Pittsburgh, Pennsylvania, Aug. 18-20, 1982, pp. 314-318.
C. Batini and M. Lenzerini, "A Comparative Analysis of Methodologies for Database Schema Integration," ACM Computing Surveys, vol. 18, No. 4, Dec. 1986, pp. 323-364.
D. Heimbinger and D. McLeod, "A Federated Architecture for Information Management," ACM Transaction on Office Information Systems, vol. 3, No. 3, Jul. 1985, pp. 253-278.
M. Stefik, D. Bobrow, S. Mittal and L. Conway, "Knowledge Programming in Loops: Report on an Experimental Course," The A1 Magazine, Fall 1983, pp. 3-13.
D. Tsichritzis, "Form Management," Communications of the ACM, Jul. 1982, vol. 25, No. 7, pp. 453-478.
H. Putnam, "The Meaning of `Meaning`," In Language, Mind, and Knowledge, ed. by K. Gunderson, Minneapolis: Univ. of Minnesota Press, 1975, pp. 131-193.
D. Lowe, "Co-operative Structuring of Information: the Representation of Reasoning and Debate," In the. J. Man-Machine Studies (1985), vol. 23, No. 1, pp. 97-111.
F. Halasz, "Reflections on Notecards: Seven Issues for the Next Generation of Hypermedia Systems," Comm. of the ACM, Jul. 1988, vol. 31, No. 7, pp. 836-852.
M. Stefik, G. Foster, D. Bobrow, K. Kahn, S. Lanning and L. Suchman, "Beyond the Chalkboard: Computer Support for Collaboration and Problem Solving in Meetings," Comm. of the ACM, Jan. 1987, vol. 30, No. 1, pp. 32-47.
J. Guttag, "Abstract Data Types and the Development of Data Structures," Comm. of the ACM, Jun. 1977, vol. 20, No. 6, pp. 396-404.
G.W. Furnas, "Generalized Fisheye Views," Proceedings of the ACM CHI Conf. on Human Factors in Computing Systems, Boston, MA, Apr. 1986, pp. 16-22.
G.R. Barber, "Office Semantics," Ph.D. thesis, Massachusetts Institute of Technology, 1982.
F.A. Turbak, "Grasp: A Visible and Manipulable Model for Procedural Programs," Unpublished M.S. Thesis, Dept. of Electrical Eng. and Computer Science, Mass. Inst. of Technology, (1986).
J. Lee, "Knowledge Base Integration: What Can we Learn from Database Integration Research,?" MIT Al Memo 1011, 1988.
J. Lee and T. Malone, "Translating Type Hierarchies: Framework Analysis and A Proposal," Sloan Work Paper No. 1974-88, M.I.T., Jan. 1988.
Lai, Malone, and Yu; Object Lens: A "Spreadsheet" for Cooperative Work, pp. 1-28, figs. 1-11, Sep. 26, 1988, Proceedings of the ACM Conference on Computer-Supported Cooperative Work, Portland, OR, Sep. 26-28, 1988.
Borland International; Reflex Plus: The Database Manager--User's Guide, pp. i-142, 1987, Scotts Valley, CA.
Ansa Software, Introduction to PARADOX, 1987, Scotts Valley, CA.
Enable Software Higgins Group; Higgins Gropu Productivity Software -Reference Manual, 1989, Alameda, CA.
Edited by Zdonik and Daier; Readings in Object-Oriented Database Systems, Chap. 3, pp. 163-226; Chap. 8, pp. 519-566, 1990, San Mateo, CA.
Stefik and Bobrow; Object-Oriented Programming: Themes and Variations, The AI Magazine, pp. 40-62, 1984, Palo Atlo, CA.
Hewlett Packard; HP NewWave User's Guide, 1990.
Fikes and Kehler; The Role of Frame-Based Representation in Reasoning, Communications of the ACM--vol. 28, No. 9, pp. 904-920, Sep. 1985.
Goodman; The Complete HyperCard Handbook, Sep. 1987, New York, NY.
Conklin and Begeman; glBIB: A Hypertext Tool for Exploratory Policy Discussion, MCC Software Technology Program, pp. 1-22, 1988, Austin, TX.
Fikes; Integrating Hypertext and Frame-Based Domain Models, May 1988, Menlo Park, CA.
Delisle and Schwartz; Contexts--A Partitioning Concept for Hypertext, ACM Transactions on Office Information Systems, vol. 5, No. 2, pp. 168-186, Apr. 1987, Beaverton, OR.
Malone; What is Coordination Theory?, Feb. 19, 1988, Cambridge, MA.
Mackay, Malone, Crowston, Rao, Rosenblitt, and Card; How do Experienced Information Lens Users Use Rules?, Oct. 1988, Palo Alto, CA.
Malone, Grant, Turbak, Brobst, and Cohen; Intelligent Information-Sharing Systems, Communications of the ACM, vol. 30, pp. 390-402, May 1987.
Ashton-Tate; Learning and Using dBase III Plus, 1985.
Claris Corporation; MacProject II User's Guide, 1989, Santa Clara, CA.
Lotus Development Corporation; Lotus Agenta User's Guide, 1988, Cambridge, MA.
Crowston and Malone; Computational Agents to Support Cooperative Work, Mar. 1988, Cambridge, MA.
Lee and Malone; Partially Shared Views A Scheme for Communicating among Groups that Use Different Type Hierarchies, Mar. 1988, Cambridge, MA.
Apple Computer, Inc.; Human Interface Guidelines: The Apple Desktop Interface, 1987, Cupertino, CA.
Malone, T.W., et al., "Intelligent Information Sharing Systems," 30 Communications of the ACM, No. 5, (May 1987) 390-402.
Malone, T.W., et al., "Semistructured Messages Are Surprisingly Useful for Computer-Supported Coordination," 5 ACM Transactions on Office Information Systems, No. 2 (Apr. 1987), 115-131.
Kee Manual.
Borland, "dBase for Windows", 1984, pp. 113-167.

Abstract

Description