Latent user communities

1. A method implemented by at least one server computer, comprising:

providing, over the Internet, access to a plurality of topics, wherein each topic includes, and further provides access to, a plurality of posted items;

recording interaction data for the plurality of topics, the interaction data identifying user activity occurring within each of the plurality of topics;

analyzing the interaction data to identify clusters of topics that exhibit similar activity-behavioral patterns, wherein analyzing the interaction data includes,

determining a covariance amongst the plurality of topics using the interaction data,

generating a sparse graph using the determined covariance, and

processing the sparse graph to identify the clusters of topics that exhibit similar behavioral patterns;

for each cluster of topics, generating a community that includes topics in the cluster; and

providing, over the Internet, access to the communities, wherein accessing a given community further provides access to topics included in that community, which further provide access to posted items that are included in the topics within that community.

2. The method of claim 1 ,

wherein determining the covariance amongst the plurality of topics includes generating a covariance matrix using the interaction data;

wherein generating the sparse graph includes applying a graphical lasso to the covariance matrix;

wherein processing the sparse graph includes applying a stochastic block model to the sparse graph.

3. The method of claim 1 , wherein the interaction data defines a time-dependent metric for user activity within the plurality of topics, and wherein a similar behavioral pattern for a given cluster of topics is defined by similar changes in the time-dependent metric occurring over time for the given cluster of topics.

4. The method of claim 3 , wherein the interaction data includes a number of one or more of active users, posted items, comments, votes, positive endorsements, negative endorsements, abuse flags, moderator actions.

5. The method of claim 1 , wherein providing access to a given topic includes providing an interface for posting items within the given topic, and for interacting with posted items in the given topic, and wherein the interaction data includes data obtained via the interface.

6. The method of claim 1 , wherein the plurality of posted items include news articles.

7. A method implemented by at least one server computer, comprising:

providing, over the Internet, access to a plurality of topics, wherein each topic includes, and further provides access to, a plurality of posted items;

recording interaction data for the plurality of topics, the interaction data identifying user activity occurring within each of the plurality of topics;

analyzing the interaction data to identify clusters of topics that exhibit similar behavioral patterns, wherein analyzing the interaction data includes,

determining a covariance amongst the plurality of topics using the interaction data,

generating a sparse graph using the determined covariance, and

processing the sparse graph to identify the clusters of topics that exhibit similar behavioral patterns;

using the identified clusters of topics to identify topics for recommendation to a user; and

presenting, over the Internet, the topics identified for recommendation in a session for the user.

8. The method of claim 7 , wherein the topics for recommendation are identified based on identifying a topic to which the user has subscribed, and identifying further topics that are within a same cluster as the topic to which the user has subscribed.

9. The method of claim 7 ,

wherein determining the covariance amongst the plurality of topics includes generating a covariance matrix using the interaction data;

wherein generating the sparse graph includes applying a graphical lasso to the covariance matrix;

wherein processing the sparse graph includes applying a stochastic block model to the sparse graph.

10. The method of claim 7 , wherein the interaction data defines a time-dependent metric for user activity within the plurality of topics, and wherein a similar behavioral pattern for a given cluster of topics is defined by similar changes in the time-dependent metric occurring over time for the given cluster of topics.

11. The method of claim 10 , wherein the interaction data includes a number of one or more of active users, posted items, comments, votes, positive endorsements, negative endorsements, flags, moderator actions.

12. The method of claim 7 , wherein providing access to a given topic includes providing an interface for posting items within the given topic, and for interacting with posted items in the given topic, and wherein the interaction data includes data obtained via the interface.

13. The method of claim 7 , wherein the plurality of posted items include news articles.

14. A method implemented by at least one server computer, comprising:

providing, over the Internet, access to a plurality of topics, wherein each topic includes, and further provides access to, a plurality of posted items;

recording interaction data for the plurality of topics, the interaction data identifying user activity occurring within each of the plurality of topics;

analyzing the interaction data to identify clusters of topics that exhibit similar behavioral patterns, wherein analyzing the interaction data includes,

determining a covariance amongst the plurality of topics using the interaction data,

generating a sparse graph using the determined covariance, and

processing the sparse graph to identify the clusters of topics that exhibit similar behavioral patterns; and

using the identified clusters to recommend specific user-posted items.

15. The method of claim 14 , wherein using the identified clusters to recommend specific user-posted items further includes, identifying a first topic to which a user has subscribed, identifying a second topic that is within a same cluster as the first topic, and recommending in a session of the user a posted item from the second topic.

16. The method of claim 14 ,

wherein determining the covariance amongst the plurality of topics includes generating a covariance matrix using the interaction data;

wherein generating the sparse graph includes applying a graphical lasso to the covariance matrix;

wherein processing the sparse graph includes applying a stochastic block model to the sparse graph.

17. The method of claim 14 , wherein the interaction data defines a time-dependent metric for user activity within the plurality of topics, and wherein a similar behavioral pattern for a given cluster of topics is defined by similar changes in the time-dependent metric occurring over time for the given cluster of topics.