Headline posting algorithm

[0002] The present invention relates to data management and transformation. More specifically, the present invention relates to the method of providing personalized data to the user in accordance with the user's profile using assigned parameters to determine which data is delivered to the particular user.

[0003] The World Wide Web is a hypertext based Internet information service. The idea behind the hypertext is that instead of reading text in a rigid linear structure (such as a book), a user can skip from one point to another. It is possible to get more information on the topic, go back, jump to other topics, and navigate through the text.

[0004] Information takes up a great deal of space, particularly when it includes images and multimedia capabilities. The Web is successful in providing so much information because that information is distributed globally across thousands of Web sites, each of which contributes the space for the information it publishes.

[0005] One distinctive feature of the Web is that it is interactive. Interactivity is the ability to “talk back” to the Web server. More traditional media such as television are not interactive at all. Other than changing a channel the viewer has little control over the content. The Web is inherently interactive: the act of selecting a link and jumping to another Web page to go somewhere else on the Web is a form of interactivity. In addition to this simple interactivity, however, pages can be designed to contain interactive forms that readers can fill out. Forms can contain text-entry areas, radio buttons, or simple menus of items. When the form is submitted, the information readers type is sent back to the server where pages originated. Despite this interactivity, however, users are still unable to limit the stream of information they are getting to the information they actually want to see.

[0006] One of the biggest problems people encounter when they explore the Web is finding information they are particularly looking for. Many Web sites providing a huge number of bytes of information are now equipped with global search engines, which search through the content of various databases looking for words that match user's query. However, this method produces a large number of irrelevant results. Consequently, the user spends a lot of time sorting through the data looking for the desired information.

[0007] Such time consuming process is especially undesirable when the searcher is looking for health related information. There are many web sites that offer information about new developments in medicine and health. It is also especially important for this type of information to be tailored to the specific needs of individual users. However, at the present time there is no efficient way of searching for specific information that the user needs or desires.

[0008] It is an object of the present invention to provide a data management system that will allow Web surfers to receive information in accordance with their interests.

[0009] It is another object of the present invention to provide an interactive Internet tool that will automatically deliver personalized information to each user.

[0010] It is a further object of the present invention to provide an interactive data management Internet tool that will minimize content searching time.

[0011] Other objects, advantages and features of this invention will be more apparent hereinafter.

[0012] The program uses the topic interest flags to determine which articles and other customizable elements of the database are shown on a particular member's home page. The headline posting algorithm first checks the flags on every available item against the member's interests, recorded in his or her user profile, then weighs the importance, rarity and significance of each match it has found, and finally displays the three highest-scoring items on the member's home page in descending order of value.

[0013] A full understanding of the invention can be gained from the following description of the preferred embodiment when read in conjunction with the accompanying drawings in which:

[0014] FIG. 1 is a schematic block diagram of the steps to be performed by the posting algorithm while selecting the headlines to be posted on the user's home page;

[0015] FIG. 2 is a schematic block diagram of the steps to be performed by the posting algorithm within the block 10 in FIG. 1, while searching for the first headline;

[0016] FIG. 3 is a schematic block diagram of the steps to be performed by the posting algorithm within blocks 20 and 30 in FIG. 1, while searching for the second and the third headline;

[0017] FIG. 4 is a registration screen which allows each user to select among the short list of very broad topics;

[0018] FIGS. 5-16 are registration screens which are presented to each user who selects the topics shown in FIG. 4;

[0019] FIG. 17 is the screen featuring three headlines on the Member's Home-Page selected for one particular user;

[0020] FIG. 18 is the screen showing the “Whose Body Do You Have” button;

[0021] FIG. 19 is the screen showing the dialog box that the user has to fill out to allow the algorithm to find the user's “body twin”;

[0022] FIG. 20 is the screen showing the user his or hers “body twin;”

[0023] FIG. 21 is a partial list of topic flags used by the algorithm to identify user's interests;

[0024] FIG. 22 is the E-chart used to perform the Home Eye Exam;

[0025] FIG. 23 is a dialog box which is used for accurate scaling of the E-chart;

[0026] FIG. 24 is a 7-minute check-up opening screen;

[0027] FIGS. 25-30 are check-up screens which are presented to each user who selects the 7-minute check-up from FIG. 24;

[0028] FIGS. 31-34 are check-up results screens; and

[0029] FIG. 35 is the Amsler Greed used to test for macular degeneration.

[0030] The following Detailed Description describes the invention in connection with the headline posting algorithm already partially implemented at the HealthScout.com, the web site containing health related materials, and its predecessor RemedyOnline.com. It is to be understood, however, that the described algorithm can be implemented with any Internet database regardless of its content.

[0031] When a new member signs on he/she is prompted through a series of screens to develop a member profile. The profile can be simplified, such as selecting several topics of interest from the list pictured in FIG. 4, or a detailed one. To compile a detailed profile, each item in the FIG. 4 is linked to a screen containing several subitems. These subitems are partially shown in FIGS. 5-16. User selects the topics of interest by clicking on the corresponding subitem. Each clicked topic is then stored in the memory of the server as a topic flag, thereby creating a personalized member profile.

[0032] Alternatively, the personalized profile can be created from the answers to and results of a health-risk assessment questionnaire, the “7-Minute Checkup”, which can be performed at the time of registration. The questionnaire algorithm compares user's answers with stored statistical data and calculates this user's risk with regard to specific health problems. Topic flags associated with the found health problems will be stored in the member's profile. If the check-up indicates an emergency or if the algorithm described hereinafter finds an item that is very important to the member health, the algorithm may sent an alert message to the member via an e-mail or fax. Alternatively, the program may put the alert message to a server's operator instructing the operator to call the member. For these circumstances there can also be a doctor “on-call” capable of helping the member to take proper actions.

[0033] Every element of the database, such as articles, newsbytes, features, etc., is evaluated for its content before being stored on the server. During the evaluation such element is assigned a number of flags corresponding to its content. For example, an article about estrogen replacement therapy will get flags ESTR (for estrogen), WMEN (for women's subject matter), BRS (for possible connection to breast cancer), FHOR (for female hormones) and so forth, depending on the content. The sample list of flags with corresponding topics is shown in FIG. 21. There are more than 700 flags that are currently used at HealthScout.com.

[0034] Each topic (i.e. flag) is graded (evaluated) on the basis of two factors: importance factor (IF) and rarity factor (RF). The IF is a number from 1 to 9 indicating how important that topic is to the average person's health. The RF is a number from 1 to 9 indicating how rarely the topic would be of personal interest among the general population. Since these criteria are uniform for all data corresponding to one topic, individual elements of data are not reviewed to determine individual IF and RF, rather these factors are automatically assigned to all elements within a particular topic. Other factors may be developed as well as rankings for said factors.

[0035] In contrast, a third criterion for analysis—significance factor (SF)—is assigned to a particular element of data. The significance factor (SF) is a number from 1 to 9 indicating how significant is a particular newsbyte or other element itself. The editors assign this factor to each story when it is posted.

[0036] Each topic also has a “day rating”. The “day rating” refers to the day of the week on which a particular topic (flag) is checked by the posting algorithm. “E” means the flag is checked every day; “0” means it is checked on the first log-on of the week; and “1” through “7” mean it is checked only on Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, and Saturday respectively.

[0037] Consequently, each element of data has four values attached to it: the IF, RF and the “day rating” corresponding to the topic, and the SF corresponding to the element itself. Each of these values can be a one-byte symbol or any other byte-value simbol to which the flag or the individual element of data within the flag can be set. “Blank” is also a valid setting in each case. These are the customizing data that the algorithm looks at when deciding which headlines to post.

[0038] Referring to FIG. 1, the process of selecting the headlines involves several steps performed by the algorithm. As shown in block 12, the first step performed by the algorithm is to determine whether this particular member has already logged on earlier on this same day. To enable this step to be performed, the server maintaines a record of the time spent on the site for each individual member. If the record shows that the member visited the site previously on the same day the algorithm will post the same headlines the member saw before, block 60.

[0039] If, however, the record shows that the member hasn't logged on earlier on this day, the algorithm will perform step 14, which is to isolate all assets on the site that the member has not yet seen (i.e. clicked on). The step is done to ensure that the member does not see the same element of data twice.

[0040] Using the same record, the algorithm next determines whether the member visited the site on an earlier day in this calendar week, block 16. If he has not, the algorithm will isolate all of the remaining assets whose “day ratings” are either E or 0, block 18. Those are usually the most important or most interesting topics for this particular calendar week. Otherwise (block 22), the algorithm will determine the today's date, using the internal calendar, and isolate all of the remaining assets whose day flags are either E or 1 if today is Sunday (blocks 24 and 26), E or 2 if today is Monday (blocks 28 and 32), E or 3 if today is Tuesday (blocks 34 and 36), E or 4 if today is Wednesday (blocks 38 and 40), E or 5 if today is Thursday (blocks 42 and 44), E or 6 if today is Friday (blocks 46 and 48), or E or 7 if today is Saturday (blocks 50 and 52). Out of the isolated items, the algorithm will pick the individual headlines for the member. The above described series of steps, designated as Step B in FIG. 1, can be skipped altogether if capabilities of the server allow for searching the entire database.

[0041] For a more convenient presentation, the steps to be performed by the algorithm within blocks 10, 20 and 30 in FIG. 1, are shown separately in FIGS. 2 and 3. FIG. 2 shows steps to be performed in determination of the first headline to be posted. Finding of the second and third headline will involve some of the same steps, grouped together as Step A in FIGS. 2 and 3, and also some additional steps, as shown in FIG. 3.

[0042] As shown in FIG. 2 to determine which headline should be posted first, the algorithm will first find items within the database which have the most flags matching the flags in the member's profile, block 62. If there is only one such item, it will be immediately selected and posted as the first headline, blocks 72 and 54 (FIG. 1) respectively. If, however, there is a tie among several items, the algorithm will multiply the IF, RF and SF for each item and compare the products of the multiplication (designated as F1 in block 64). The algorithm will double the result if the item is a newsbyte. The algorithm will choose the item that yields the higher product, block 66. It is possible to change the above described formula in accordance with specific needs of the server.

[0043] If there is still a tie after block 66, the algorithm will compare the posting dates for each item and choose the one whose posting date is later, block 68. If there is still a tie, the program will pick the headline at random, block 70. The selected item is then displayed as the member's first headline, block 54, FIG. 1.

[0044] All of the steps described above for selecting the first headline, except block 72, will be repeated by the algorithm in the process of determining the second and third headline, blocks 20 and 30 respectively in FIG. 1. Therefore, for simplicity they are grouped together in FIGS. 2 and 3 as Step A, block 74. After finding the first headline the algorithm will perform the Step A again to find the next most valuable item. If it is a newsbyte or a feature, the program will post it as the second headline, blocks 76 and 56. If it is not a newsbyte or a feature, it will be posted only if it is not the same kind of item as the top headline, block 78. For example, the algorithm will not post two drug reports. If the item is of the same kind, the algorithm will perform Step A again to determine the next most valuable item, block 80, and repeat steps in blocks 76 and 78 until it will find a postable item. Found item will be displayed as the second headline, block 56. To find third headline, all of the steps described with respect to the second headline will be repeated and the result posted for the member (block 58). The personalized screen with headlines selected for the particular member under the title “Your Headlines” is shown in FIG. 17.

[0045] As a result of the described data manipulation, the user will receive a limited number of data elements which are specifically tailored to this user's interests and needs. The number of displayed items may vary, however, the algorithm for selecting additional headlines will be the same as for selecting the second and the third headlines above.

[0046] There are additional ways for the algorithm to select the headlines. On each article detail page, beside every article there is a “Rate This Article” button linked to a dialog box that invites the member to rate the article for usefulness, ease of reading and overall quality. The data collected from members in this dialog box will be used to give the algorithm a deeper understanding of members' preferences and enable it to serve up more welcome headlines. When the algorithm faces a choice between two articles that would otherwise score equally well against the user's profile, for example, it will automatically prefer the article that has scored higher with other users.

[0047] The server is keeping a record of all the articles that each member clicks on and presumably reads. When the algorithm faces a choice between two articles that would otherwise score equally well against the user's profile, it will automatically prefer the article whose flag set most closely matches those of previous articles he has clicked on.

[0048] It is also possible to apply an objective test (such as Gunning's Fog Index or the Flesch Index) to measure the readability of each article in the database. Members will occasionally use “Rate This Article” to express how easy they found an article to read. By matching a member's readability ratings with the objective ratings of the same articles, the algorithm will arrive at a preferred index for each member. Whenever the algorithm faces a choice between two articles that would otherwise score equally well against the user's profile, it will automatically prefer the article whose readability index most closely matches that preferred by the member.

[0049] It is also possible to add a special flag to each article indicating how knowledgeable about its main topic a reader would have to be in order to find the article too elementary and repetitive of facts that he or she already knows. When the algorithm looks at all of the new (never clicked on) items that would otherwise score high enough against the user's profile to merit display in the “Your Headlines,” it will discard those that are too elementary.

[0050] In addition to the personalized Member Front Page, the server offers the Guest Front Page. This page is not customized; every user sees the same headlines. At the top of the page are today's top three headlines, which the editors will decide each day. Unlike the headlines in the Your Headlines section of the Member Home-Page, which can be newsbytes or features or any of several other customizable elements of the database, the headlines on the Guest Home Page will nearly always be newsbytes. All three headlines will have graphics attached: either icons or unique art.

[0051] Clicking either the “More News” button on either home page or the “News” or “Other News” listing on the guest or member navigation-bar, respectively, will open the Week's News Page. There all of the week's newsbytes will displayed, grouped by posting date (with a header listing the date) and alphabetized by headline within each posting date. These links may have no art attached.

[0052] Each week, editors may decide which two or more features should receive prominent display at the top of the Feature Index Page. Those two may have art attached. Underneath them will come a list of all remaining features on the site, grouped in an alphabetical list of feature categories and arranged alphabetically by headline within each category. The number and names of the categories will change over time as new ones are added, old ones are removed, new and old ones are consolidated, etc. Headlines in this part of the page may have no art attached, except for “New” bursts that the editors can attach by activating a flag and remove by switching the flag off.

[0053] Besides filtering the contents of one database located on the server, the algorithm can also look at the articles on any other Web site chosen by the member and pick items most appropriate to the member's personal interests. It will do this by applying both keyword and concept analysis to the target site's articles in order to infer which of the topic flags they should have attached to them, and then using the previously described parts of the algorithm to decide which articles to display in personal headlines section.

[0054] It is also possible to make the algorithm continually comb the entire Web for the articles of greatest personal interest to the member. The algorithm will then combine “spidering” technology, currently used by the global search engine software, with keyword and concept analysis. At this point, the server will be functioning as an autonomous personal agent, applying artificial intelligence to select the most pertinent items from all over the Web, without any real-time direction from its user.

[0055] In addition to the above described services, the server can also store medical encyclopedias and search them at member's request. It can also provide information about physicians and drugs, as well as links to other health information resources on the Internet.

[0056] As entertainment, the server may also find a celebrity “body twin” for the user. When the user will click on “Whose Body Do You Have” button, FIG. 18, the dialog screen will appear asking the user to enter the gender, date of birth, weight and height and then to click on “Match Me” button, FIG. 19. To find the “body twin” the algorithm will calculate the user's BMI and round it to the nearest whole number. BMI=(weight in kilograms)/(height in meters)². It will then isolate every celebrity in the database who has the same gender and body type as the user and, from that group, isolate all who share the user's BMI. If none do, it will isolate the group whose BMI is closest to the user's. If there's a tie, the algorithm will isolate the larger of the two groups. The next step is to isolate from that group all who share the user's birthday (irrespective of birth year). If there are none who share the user's birthday, the algorithm will then isolate all who share the user's zodiacal sign. If there are none who share the user's zodiacal sign, it will then keep this whole group. From the last resulting group, the algorithm will pick the celebrity whose birth year is closest to the user's. If there's a tie, it will pick the one whose first name is first in the database alphabetically. The found celebrity's name and biographical information will appear on the user's screen as shown in FIG. 20.

[0057] To provide a more complete health service to the user, it is also possible to implement a Home Eye Exam on the same server by including a link to an Eye Exam page on the Member Front Page. During the Home Eye Exam the user will be able to determine whether his or her vision has become worse since the last check-up and whether the new prescription is needed. The actual eye-exam is performed using a regular E-chart, shown in FIG. 22. However, to provide accurate results the E-chart must first be adapted to the user's screen. Therefore, as the user will click on Home Eye Exam button, a black square and a dialog box shown in FIG. 23 will appear on the screen. Measurements of the square on the user's screen will be used by the algorithm to determine proper parameters of the E-chart which will be displayed on the same screen. The square is sized so that an average monitor, probably a 14-inch with 640-by-480 resolution, will display the square as one inch wide. This should make the largest likely square (on a 20-inch monitor at 640-by-480 resolution) just over 1⅜ inches wide and the smallest likely (on a 14-inch at 1024-by-768) about ⅝ inches wide.

[0058] Meanwhile, the “E” chart is scaled for viewing at 10 feet when the black square is one inch wide. This means that when the black square is an inch wide on the screen, the Es on the chart should be of the following widths on the screen: For 20/100 vision, {fraction (56/64)} or ⅞ inch; for 20/70 vision, {fraction (39/64)} inch; for 20/40 vision, {fraction (22/64)} or {fraction (11/32)} inch; for 20/30 vision, {fraction (16/64)} or ¼ inch; for 20/20 vision, {fraction (11/64)} inch; for 20/15 vision, {fraction (8/64)} or ⅛ inch. In addition, if there is room on the screen it is possible also to add a set of Es for 20/10 vision. These would each be {fraction (6/64)} inch wide.

[0059] Finally, all the viewing distances for the test should be made directly proportional to the size of the black square on the user's monitor. Clicking the “1 inch” radio button should insert “10 feet 0 inches” in the boxes 82 in FIG. 23; clicking “⅝” should insert “6 feet 3 inches;” clicking “¾” should insert “7 feet 6 inches;” clicking “⅞” should insert “8 feet 9 inches;” clicking “1⅛” should insert “11 feet 3 inches;” clicking “1¼” should insert “12 feet 6 inches;” clicking “1⅜” should insert “13 feet 9 inches;” and clicking “1½” should insert “15 feet 0 inches.”

[0060] These preparations will insure that the smallest group of Es on the E-chart in which the user can orient at least half of the Es will indicate the sharpness of his or her vision. The procedure for performing the eye-exam is known to most users and will appear on the users' screens for their reference.

[0061] The E-chart test described above is typically used to find problems with distance vision. The same preparations, however, can be used to find other vision problems such as presbyopia, macular degeneration or amblyopia. The same E-chart can be used to test for amblyopia. The test called the Amsler Grid, shown in FIG. 35, is typically used to detect macular degeneration. Although it is shown in black-on-white, the Amsler Grid should be displayed on the user's screen as white-on-black, and the entire background around the grid should be black. This is to make the test more accurate for people who have cataracts, by lessening the effects of glare. On the user's monitor, the Amsler Grid should be about four inches square. This means there should be four different grids: one to be displayed if the user has clicked on “⅝” for the width of the black square on page 2; the others to be displayed if the user has clicked on “⅞,” “1⅛” or “1⅜.” To test the user for presbyopia, the algorithm can direct the user to read a passage on the screen and see if it causes a strain in the eyes. The passage should be rendered in a type size whose capitals are 0.13 times the height that was determined for the Es in the 20/30 group.

[0062] Having described this invention with regard to specific embodiments, it is to be understood that the description is not meant as a limitation since further variations or modifications may be apparent or may suggest themselves to those skilled in the art. It is intended that the present application cover such variations and modifications as fall within the scope of the appended claims.