System for adjusting hypertext links with weighed user goals and activities

• US 5,446,891 A
• Filed: 11/02/1994
• Issued: 08/29/1995
• Est. Priority Date: 02/26/1992
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Chen, James R., Fenwick, Justine R., Wolff, Gregory J., Kaplan, Craig A., Fallside, David C., Forcier, Mitchell D.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): AMSBURY, WAYNE P
• Application Number:
  US08/333,082
• Time in Patent Office: 300 Days
• Field of Search: 395/600
• US Class Current: 1/1
• CPC Class Codes: G06F 16/94 Hypermedia Hyperlinking G06F40/134
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A smart hypermedia system that acquires user characteristics either directly or inferentially. Simple associative networks serve to model user profiles, including relationships between user goals and the hypermedia information nodes. Hypermedia links to other nodes are recommended by ranking a link list in an order that depends on one or more user profiles containing information relating to users'"'"' goals and interests. Users can teach the system directly by rearranging the order of suggested links on the list. The system can also learn indirectly by observing how long and in what sequence the user views each hypermedia information node. User profiles can be combined to form group profiles and may be dynamically and continuously updated to form an adaptive system profile. The two system learning modes may be simultaneous or disjoint.

Claims

1. A computer-implemented hypertext system comprising:

display means for displaying images and text;

keyboard means for accepting user commands;

memory means for storing a plurality of data objects including user goal objects and text panel objects organized to form at least one hypertext object;

profile storage means for storing at least one link profile each associated with a specific set of one or more users and containing a plurality of link vectors each containing a plurality of numerical link-weights each representing an associated user activity relationship between two said data objects; and

advisor means for creating an ordered list of one or more said data objects responsive to user input activity, said list being ordered in accordance with the relative values of said numerical link-weights representing said user input activity.

2. The system of claim 1 wherein said advisor means further comprises:

context definer means for selecting a link vector from a user profile responsive to said user input activity.

3. The system of claim 2 wherein said advisor means further comprises:

clock means for indicating elapsed time; and

autolearner means for updating a user profile responsive to the data object selection activity and said elapsed time associated with said user input activity.

4. The system of claim 3 wherein said advisor means further comprises:

profile processor means for combining a first plurality of said link profiles to form one or more new link profiles each associated with a combined set of users including users associated with link profiles from said first link profile plurality.

5. The system of claim 1 wherein said advisor means further comprises:

clock means for indicating elapsed time; and

autolearner means, within said advisor means for updating a user profile responsive to the data object selection and viewing time associated with said user input activity.

6. The system of claim 1 wherein said advisor means further comprises:

profile processor means for combining a first plurality of said link profiles to form one or more new link profiles each associated with a combined set of users including users associated with link profiles from said first link profile plurality.

16. The method of claim 1 further comprising the steps of:

(e) modifying said first link profile responsive to user activity at said keyboard means; and

(f) storing said modified first link profile in said memory means.

7. A method for operating a computer-implemented object-oriented hypertext system having a plurality of link profiles each associated with a specific set of one or more users and containing a plurality of link vectors each containing a plurality of numerical link-weights, each said link-weight representing an associated user activity relationship between one of a plurality of data objects and another one of said plurality of data objects, said plurality of data objects including text panel objects and user goal objects, said system also having display means for displaying images and text, keyboard means for accepting user commands, and at least one memory means for storing said data objects and link profiles, said method comprising the steps of:

(a) selecting a first link profile having a first link vector corresponding to a first data object, wherein said first data object is either a text panel object or a user goal object;

(b) displaying at least part of said first data object on said display means responsive to said keyboard means;

(c) comparing the link-weights within said first link vector to find a maximum link-weight value; and

(d) displaying on said display means at least part of a second data object corresponding to said maximum-valued link-weight, wherein said second data object is a text panel object.

8. The method of claim 7 further comprising the steps of:

(e) sorting into numerical order said link-weights in said first link vector; and

(f) displaying at least one identifier from each of one or more said data objects and the associated link-weight ordering for each said identifier displayed.

9. The method of claim 8 wherein said displaying step (f) further the comprises the step of:

(f.1) displaying said one or more displayed identifiers in order of the value of said associated link-weight.

10. The method of claim 9 further comprising the steps of:

(g) comparing said associated link-weight values in said first link vector with a threshold value; and

(h) displaying said identifiers for data objects associated with link-weight values greater than said threshold value.

11. The method of claim 8 further comprising the step of:

(g) changing the link-weight value associated with a selected data object responsive to user input activity at said keyboard means, whereby the sort order of said displayed identifiers is changed responsive to said user input activity.

12. The method of claim 7 wherein said selecting step (a) further comprises the steps of:

(a.1) merging a first plurality of said link profiles to form a combined link profile associated with a combined set of users including each user associated with a link profile from said first link profile plurality; and

(a.2) selecting said combined link profile to be said first link profile.

13. The method of claim 12 wherein said merging step (a.1) further comprises the steps of:

(a.1.1) calculating the link-weight values for a system link profile R by merging the link-weight values of a new link profile R_n with the corresponding link-weight values of an existing system link profile R_e according to the formula, R=(R_e *(K-1)+R_n)/K, where K is a predetermined constant; and

(a.1.2) storing said link-weight values for said system link profile R in said memory means.

14. The method of claim 13 wherein said calculating step (a.1.1) further comprises the step of:

(a.1.1.1) initializing said link-weight values in said system link profile R by merging a new link profile R_n to an existing system link profile Re according to the formula, R=(R_e *N+R_n)/(N+1), where N<

K and N is the number of profiles that have been previously merged to form said existing system link profile R_e.

15. The method of claim 12 wherein said merging step (a.1) further comprises the steps of:

(a.1.1) calculating, for each link between data objects j and i, a normalized link-weight value R_ij representing the corresponding link-weight value for the merger of a set of link profiles, by first assigning a profile weight W_p to the p^th said link profile in a subset (ij) of said set and then averaging said subset (ij) according to the formula R_ij =Σ

_p (R_ijp *W_p)Σ

_p W_p, where said subset (ij) excludes all said link profiles in said set that do not contain a link-weight value relating data objects j and i; and

(a.1.2) storing said normalized link-weight values R_ij in said combined link profile in said memory means.

17. A method for operating a computer-implemented object-oriented hypertext system having a plurality of link profiles each associated with a specific set of one or more users and containing a plurality of link vectors each containing a plurality of numerical link-weights, each said link-weight representing an associated user activity relationship between one of a plurality of data objects and another one of said plurality of data objects, said plurality of data objects including text panel objects and user goal objects, said system also having display means for displaying images and text, keyboard means for accepting user commands, clock means for indicating elapsed time, and at least one memory means for storing said data objects and link profiles, said method comprising the steps of:

(a) selecting a first link profile having a first link vector corresponding to a first data object and a second link vector corresponding to a second data object, wherein said first data object is either a text panel object or a user goal object and said second data object is a text panel object;

(b) displaying at least part of said first data object on said display means;

(c) displaying at least part of said second data object on said display means responsive to user activity at said keyboard means;

(d) recording a start time from said clock means to begin measurement of an elapsed time interval;

(e) monitoring for a predetermined event signaling that said elapsed time interval should end;

(f) recording a stop time from said clock means to complete said measurement of said elapsed time interval;

(g) computing a view time for said second data object by subtracting said start time from said stop time;

(h) calculating a new link-weight value representing said associated user activity relationship between said first data object and said second data object, said new link-weight value being at least partly proportional to said view time; and

(i) storing said new link-weight value in said first link profile in said memory means.

18. The method of claim 17 further comprising the steps of:

(j) comparing the link-weights within said second link vector to find a maximum link-weight value; and

(k) displaying at least part of a third data object corresponding to said maximum-valued link-weight, wherein said third data object is a text panel object.

19. The method of claim 18 further comprising the steps of:

(l) sorting into numerical order said link-weights in said second link vector; and

(m) displaying an identifier for each of one or more corresponding said text panel objects, said identifier including the associated link-weight ordering for said each corresponding text panel object.

20. The method of claim 19 wherein said displaying step (m) further comprises the step of:

(m.1) displaying said identifiers in order of the value of said associated link-weight.

21. The method of claim 20 further comprising the steps of:

(n) comparing said associated link-weight values in said second link vector with a threshold value; and

(o) displaying only said identifiers for said corresponding text panel objects associated with link-weight values greater than said threshold value.

22. The method of claim 19 further comprising the steps of:

(n) modifying said first link profile responsive to user activity at said keyboard means; and

(o) storing said modified first link profile in said memory means.

23. The method of claim 17 wherein said selecting step (a) further comprises the steps of:

(a.1) merging a first plurality of said link profiles to form a combined link profile associated with a combined set of users including each user associated with a link profile from said first link profile plurality; and

(a.2) selecting said combined link profile to be said first link profile.

24. The method of claim 23 wherein said merging step (a.1) further comprises the steps of:

(a.1.1) calculating the link-weight values for a system link profile R by merging the link-weight values of a new link profile R_n with the corresponding link-weight values of an existing system link profile R_e according to the formula, R=(R_e *(K-1)+R_n)/K, where K is a predetermined constant; and

(a.1.2) storing said link-weight values for said system link profile R in said memory means.

25. The method of claim 24 wherein said calculating step (a.1.1) further comprises the step of:

(a.1.1.1) initializing said link-weight values in said system link profile R by merging a new link profile R_n to an existing system link profile R_e according to the formula, R=(R_e *N+R_n)/(N+1), where N<

K and N is the number of profiles that have been previously merged to form said existing system link profile R_e.

26. The method of claim 23 wherein said merging step (a.1) further comprises the steps of:

(a.1.1) calculating, for each link between said first and second data objects j and i, a normalized link-weight value R_ij representing the corresponding link-weight value for the merger of a set of link profiles, by first assigning a profile weight W_p to the p^th said link profile in a subset (ij) of said set and then averaging said subset (ij) according to the formula, R_ij =Σ

_p (R_ijp *W_p)/Σ

_p W_p, where said subset (ij) excludes all said link profiles in said set that do not contain a link-weight value relating said first and second data objects j and i; and

(a.1.2) storing said normalized link-weight values R_ij in said combined link profile in said memory means.

27. The method of claim 17 wherein said monitoring step (e) further comprises the step of:

(e.1) monitoring for an input signal from said keyboard means representing user activity.

28. The method of claim 17 wherein said calculating step (h) further comprises the step of:

(h.1) recalculating said new link-weight value to be at least partly inversely proportional to the quantity of data contained in said second data object.

29. A combination for use with a computer-implemented hypertext system, said combination including display means for displaying images and text and a program medium for storing program and data objects and structures comprising:

a program object structure stored in said program medium, said program object structure including a plurality of program objects including;

a program object for controlling said hypertext system including a plurality of link profiles each associated with a specific set of one or more users, each said link profile containing a plurality of link vectors each containing a plurality of numerical link-weights, each said link-weight representing an associated user activity relationship between one of a first plurality of said data objects and another one of said first plurality of data objects, said first plurality of data objects including text panel objects and user goal objects;

a display program object for controlling the display of images and text on said display means;

a program object for controlling a clock means for indicating elapsed time;

a program object for controlling a selection of a first link profile having a first link vector corresponding to a first data object and a second link vector corresponding to a second data object, wherein said first data object is either a text panel object or a user goal object and said second data object is a text panel object; and

a plurality of said program objects including an object for displaying at least part of said first data object on said display means, an object for displaying at least part of said second data object on said display means responsive to user activity at said keyboard means, an object for recording a start time from said clock means to begin measurement of an elapsed time interval, an object for monitoring for a predetermined event signaling that said elapsed time interval should end, an object for recording a stop time from said clock means to complete said measurement of said elapsed time interval, an object for computing a view time for said second data object by subtracting said start time from said stop time, an object for calculating a new link-weight value representing said associated user activity relationship between said first data object and said second data object, wherein said new link-weight value is at least partly proportional to said view time, and an object for controlling the storing of said new link-weight value in said first link profile data object.