METHOD AND APPARATUS FOR INTERACTING WITH INFORMATION DISTRIBUTION SYSTEM

1 . A method of interacting with an information system, comprising:

employing a hardware processor to execute processor-readable instructions to perform processes of:

acquiring contents of a plurality of articles accessible through the information system;

determining pairwise inter-article content similarity;

tracking a plurality of users of the information system to identify pairwise article successions, wherein a pairwise article succession comprises two articles accessed by a same user;

determining composite pairwise affinity levels of said plurality of articles according to:

respective inter-article content similarity;

types of tracked users effecting said pairwise article successions; and

pairwise frequency of article successions; and

determining for a designated article of said plurality of articles a preferred succeeding article according to said composite pairwise affinity levels.

2 . The method of claim 1 further comprising communicating an identifier of said preferred succeeding article to a user accessing the designated article.

3 . The method of claim 1 or 2 further comprising:

segmenting said plurality of users into a plurality of clusters according to a predefined criterion; and

determining said types of tracked users as identifiers of respective clusters to which said tracked user belong.

4 . The method of claim 1 further comprising associating each tracked user with a respective group of users and a level of significance within said respective group of users, said types indicating for said each tracked user:

a group of users to which said each tracked user belongs; and

a respective level of significance.

5 . The method of claim 1 further comprising:

detecting an access transition to a subsequent article following said communicating; and

updating a measure of effective recommendations subject to a determination that said subsequent article is the preferred succeeding article.

6 . The method of claim 1 further comprising:

detecting an access transition to a subsequent article following said communicating;

determining a first composite affinity level of said designated article to said preferred succeeding article;

determining a second composite affinity level of said designated article to said subsequent article;

updating discrepancy statistics based on comparing said first composite affinity level and said second composite affinity level.

7 . The method of claim 1 further comprising:

ranking directed article pairs originating from said designated article according to composite pairwise affinity levels;

designating a predefined number of directed article pairs as candidate directed article pairs according to said ranking; and

selecting said preferred succeeding article from among said candidate directed article pairs.

8 . The method of claim 7 wherein said selecting comprises using a randomly sequenced round robin process weighted according to composite pairwise affinity levels of said candidate directed article pairs.

9 . The method of claim 7 further comprising excluding a directed article pair of inter-article content similarity exceeding a predefined threshold.

10 . The method of claim 1 wherein said determining pairwise inter-article content similarity comprises:

formulating word vectors, each word vector characterizing content of a respective article of said plurality of articles; and

performing pairwise comparisons of word vectors of different articles.

11 . The method of claim 1 further comprising storing in a memory device coupled to said hardware processors composite pairwise affinity levels exceeding a predefined lower bound.

12 . A method, implemented in a computing device, of interacting with an information system, the method comprising:

tracking a plurality of users accessing a plurality of articles through the information system;

determining for each tracked user:

a respective user type of a predefined plurality of user types; and

a currently accessed article;

for each article-access transition where a particular user accesses a first article then a second article:

maintaining a global measure and a user-type measure of transitions from the first article to the second article; and

determining a composite measure as a function of the global measure and the user-type measure; and

recommending a first target article to succeed said currently accessed article according to composite measures of directed article pairs originating from said currently accessed article.

13 . The method of claim 12 further comprising:

acquiring contents of said plurality of articles; and

determining pairwise content similarities of said plurality of articles.

14 . The method of claim 13 further comprising:

determining a composite affinity level for each directed pair of articles as a function of at least one of:

a respective content similarity;

a respective global measure; and

a respective user-type measure;

recommending a second target article to succeed said currently accessed article according to composite affinity levels of directed article pairs originating from said currently accessed article.

15 . The method of claim 12 further comprising:

acquiring characteristics of said plurality of users;

clustering said plurality of users into a number of clusters according to said characteristics and a predefined criterion; and

determining said user type as an identifier of a cluster to which said tracked user belongs.

16 . The method of claim 15 further comprising:

determining centroids of said plurality of clusters;

determining a centroid-proximity measure of said particular user according to proximity of said particular user to a respective centroid; and

determining said user-type measure as cumulative centroid-proximity measures of users effecting said each article-access transition.

17 . The method of claim 12 further comprising for each article of said plurality of articles, ranking each other article according to a respective composite measure to produce a respective set of ranked directed article pairs.

18 . The method of claim 17 wherein said recommending comprises:

designating at least two articles of highest ranking; and

randomly designating one of said at least two articles as said target article.

19 . The method of claim 12 further comprising updating said global measure and said user-type measure following said each article-access transition.

20 . A method of interacting with an information system comprising:

employing a computing device to implement processes of:

tracking a plurality of users accessing a plurality of articles through the information system;

determining for each tracked user:

a respective user type of a predefined plurality of user types; and

a currently accessed article;

for each article-access transition where a particular user accesses a first article then a second article:

maintaining a global measure and a user-type measure of transitions from the first article to the second article; and

acquiring contents of said plurality of articles; and

determining pairwise content similarities of said plurality of articles;

determining a composite affinity level for each directed pair of articles as a function of:

a respective content similarity;

a respective global measure; and

a respective user-type measure; and

recommending a preferred article to succeed said currently accessed article according to composite affinity levels of directed article pairs originating from said currently accessed article.

21 . An apparatus for interacting with an information system, the apparatus comprising:

a processor and a plurality of memory devices storing:

a tracking module configured to track a plurality of users accessing a plurality of articles to acquire:

contents of said plurality of articles;

characteristics of said plurality of users; and

pairwise article successions;

a module for determining pairwise content-similarity levels of said plurality of articles;

a module for dividing said plurality of users into clusters according to said characteristics;

a module for accumulating for each directed article pair of said pairwise article successions:

a gravitation measure based on a respective succession count; and

an attraction measure for each cluster of users indicating a respective cluster-specific weight;

and

a recommendation module configured to communicate to a user accessing a reference article an identifier of a preferred succeeding article determined according to said pairwise content-similarity levels, said gravitation measure, and said attraction measure.

22 . The apparatus of claim 21 wherein said recommendation module is further configured to:

determine an affinity level for each directed article pair according to respective content-similarity level, gravitation measure, and attraction measure;

sort directed article pairs originating from each article into ranks according to respective affinity levels; and

determine said preferred succeeding article according to ranks of directed article pairs originating from said reference article.

23 . The apparatus of claim 21 further comprising a module, stored in one of said memory devices, configured to:

detect from said pairwise article successions, a subsequent article accessed by said user;

and

report discrepancies of content-similarity, gravitation measure, and attraction measure between transition to said subsequent article and a transition to said preferred succeeding article.