Granted Patent B2

US 12,236,468 · App. 17/718,012 · Granted Feb 25, 2025

Inventor: Philip Joseph Renaud (Toronto, CA)

Assignee: AUDIENSE GLOBAL HOLDINGS LIMITED.

G06Q30/0631G06F16/9535G06F16/9536G06F16/9537G06F18/24137G06Q30/0201G06Q50/01

Abstract

Claims (92)

1. A method, implemented at a computing system, of guiding content selection within an information distribution system, the method comprising:

acquiring pairwise similarity levels of a plurality of articles;

continually detecting access by individual users, of a plurality of users, to respective articles through the information distribution system, and for each individual user:

determining a complementing article to succeed a respective article based on:

said pairwise similarity levels; and

a gravitation measure of article-access transitions for each directed article pair emanating from said respective article, said gravitation measure being a count of article-access transitions effected by said plurality of users;

communicating an identifier of said complementing article to said each individual user;

detecting a specific directed article-access transition from said respective article to a subsequent article;

increasing a corresponding gravitation measure;

and

adjusting said gravitation measure for said each directed article pair based on

a count of transitions from said respective article; and

a global cyclic age of said gravitation measure.

2. The method of claim 1 wherein said adjusting comprises attenuating said gravitation measure for said each directed article pair subject to at least one of:

a total number of transitions from said respective article exceeding a predefined value; and

said global cyclic age of said gravitation measure exceeding a predefined time period.

3. The method of claim 1 wherein said cyclic age of gravitational measure is determined as a time interval between a time instant of a current article-access transition and a most recent time instant at which the gravitation measure has been adjusted.

4. The method of claim 1 wherein said total number of transitions is a count of all transitions from the respective article to any other article of the plurality of articles.

5. The method of claim 1 wherein said adjusting comprises reducing said gravitation measure according to one of:

a predefined multiplier; and

a function of a value of said gravitation measure.

6. The method of claim 1 further comprising:

determining a composite affinity level of said respective article to each candidate article of a set of candidate articles of said plurality of articles as a weighted sum of:

a similarity level between said respective article and said each candidate article; and

a gravitation measure from said respective article to said each candidate article; and

selecting a candidate article of said set of candidate articles as said complementing article according to a respective composite affinity level.

7. The method of claim 1 further comprising:

obtaining a specific user type of said each individual user;

refining said determining said complementing article to be further based on an attraction measure of article-access transitions for said each directed article pair, said attraction measure being a count of article-access transitions effected by users of a same user type.

8. The method of claim 7 further comprising:

updating a corresponding attraction measure of said each directed article pair for said specific user type; and

adjusting said attraction measure for said each directed article pair for said specific user type subject to a number of transitions from said respective article for said specific user type exceeding a predefined lower bound.

9. The method of claim 7 wherein said obtaining comprises:

acquiring characteristics of individual users of said plurality of users;

segmenting said plurality of users into clusters based on said characteristics; and

assigning for each user a respective user type equivalent to an identifier a cluster to which the user belongs.

10. The method of claim 7 further comprising selecting a set of candidate articles, of said plurality of articles, to complement said respective article, with each candidate article satisfying at least one of:

a respective similarity level, between said respective article and said each candidate article, exceeding a predefined similarity lower-bound;

a respective gravitation measure, from said respective article to said each candidate article, exceeding a predefined gravitation lower bound; and

a respective attraction measure, from said respective article to said each candidate article for said specific user type, exceeding a predefined attraction lower bound.

11. The method of claim 10 further comprising:

determining a composite affinity level of said respective article to each candidate article of a set of candidate articles of said plurality of articles as a weighted sum of:

said respective similarity level;

said respective gravitation measure; and

said respective attraction measure;

ranking candidate articles of said set of candidate articles according to values of respective composite affinity levels; and

performing a process of weighted random selection of said complementing article from a prescribed number of candidate articles of highest ranking.

12. An apparatus for guiding content selection within an information distribution system comprising a pool of processors executing instructions organized into:

a tracking module configured to continually detect access by individual users, of a plurality of users, to respective articles of a plurality of articles through the information distribution system;

a recommendation module configured to:

determine a complementing article to succeed a particular article based on:

acquired pairwise similarity levels of said plurality of articles;

and

a gravitation measure of article-access transitions for each directed article pair emanating from said particular article, said gravitation measure being a count of article-access transitions effected by said plurality of users;

communicate an identifier of said complementing article to a specific user currently accessing said particular article;

a module for usage-data accruing configured to:

detect a directed article-access transition from said particular article to a subsequent article; and

increase a corresponding gravitation measure;

and

a module for usage-data weighting configured to adjust said gravitation measure for said each directed article pair based on

a count of transitions from said particular article; and

a global cyclic age of said gravitation measure.

13. The apparatus of claim 12 wherein said module for usage-data weighting attenuates said gravitation measure for said each directed article pair subject to at least one of:

a total number of transitions from said particular article exceeding a predefined value; and

said global cyclic age of said gravitation measure exceeding a predefined time period;

said total number of transitions being a count of all transitions from the particular article to any other article of the plurality of articles.

14. The apparatus of claim 13 wherein said module for usage-data weighting reduces said gravitation measure according to one of:

a predefined multiplier; and

a function of a value of said gravitation measure.

15. The apparatus of claim 12 wherein said module for usage-data weighting determines said cyclic age of gravitational measure as a time interval between a time instant of a current article-access transition and a most recent time instant at which the gravitation measure has been adjusted.

16. The apparatus of claim 12 wherein said recommendation module is further configured to:

determine a composite affinity level of said particular article to each candidate article of a set of candidate articles of said plurality of articles as a weighted sum of:

a similarity level between said particular article and said each candidate article; and

a gravitation measure from said particular article to said each candidate article; and

select a candidate article of said set of candidate articles as said complementing article according to a respective composite affinity level.

17. The apparatus of claim 12 wherein said recommendation module is further configured to:

obtain a specific user type of said specific user;

refine said determining said complementing article to be further based on an attraction measure of article-access transitions for said each directed article pair, said attraction measure being a count of article-access transitions effected by users of a same user type.

18. The apparatus of claim 17 wherein said recommendation module is further configured to:

update a corresponding attraction measure of said each directed article pair for said specific user type; and

adjust said attraction measure for said each directed article pair for said specific user type subject to a number of transitions from said particular article for said specific user type exceeding a predefined lower bound.

19. The apparatus of claim 17 wherein said recommendation module is further configured to select a set of candidate articles, of said plurality of articles, to complement said particular article, with each candidate article satisfying at least one of:

a respective similarity level, between said particular article and said each candidate article, exceeding a predefined similarity lower-bound;

a respective gravitation measure, from said particular article to said each candidate article, exceeding a predefined gravitation lower bound; and

a respective attraction measure, from said particular article to said each candidate article for said specific user type, exceeding a predefined attraction lower bound.

20. The apparatus of claim 19 wherein said recommendation module is further configured to:

determine a composite affinity level of said particular article to each candidate article of a set of candidate articles of said plurality of articles as a weighted sum of:

said respective similarity level;

said respective gravitation measure; and

said respective attraction measure;

rank candidate articles of said set of candidate articles according to values of respective composite affinity levels; and

perform a process of weighted random selection of said complementing article from a prescribed number of candidate articles of highest ranking.

Assignments (6)

Continuity (3)

References Cited (73)