Social network model for semantic processing

• US 9,183,535 B2
• Filed: 07/30/2009
• Issued: 11/10/2015
• Est. Priority Date: 07/30/2008
• Status: Active Grant

• Current Assignee: ARO, Inc. (Dibona Holding)
• Original Assignee: ARO, Inc. (Dibona Holding)
• Inventors: Eustice, Kevin Francis, Lawson, Nolan, Yetisgen-Yildiz, Meliha, Kawai, Kenji, Perkowitz, Michael, Ferrel, Patrick James, Lazarus, Jonathan D.
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): HANCHAK, WALTER P
• Application Number:
  US12/512,752
• Publication Number:
  US 20100030715A1
• Time in Patent Office: 2,294 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06Q 10/10 Office automation; Time management

Abstract

A social network model, based on data relevant to a user, is used for semantic processing to enable improved entity recognition among text accessed by the user. An entity extraction module of the server, with reference to a general training corpus, general gazetteers, user-specific gazetteers, and entity models, parses text to identify entities. The entities may be, for example, people, organizations, or locations. A social network module of the server builds the social network model implicit in the data accessed by the user. The social network model includes the relationships between entities and an indication of the strength of each relationship. The social network module is also used to disambiguate names and unify entities based on the social network model.

Claims

1. A computer-implemented method of generating a user'"'"'s social network model, the method comprising:

receiving a set of documents associated with a user;

accessing the user'"'"'s contact data, the contact data identifying a plurality of entities;

analyzing the documents, using the contact data, to identify references to entities therein;

identifying relationships among the referenced entities;

determining a strength of a first relationship between a first entity and a second entity responsive to a volume of documents in which both the first entity and the second entity appear, wherein the first and second entities are a subset of the referenced entities;

building a social network model for the user responsive to the identified relationships

among the referenced entities and the strength of the first relationship;

storing the social network model;

receiving a new document associated with the user;

identifying, in the new document, a reference to an ambiguous entity;

performing name disambiguation using the social network model to determine which of

at least two candidate entities from the social network model is an intended entity for the ambiguous entity;

identifying other entities referenced by the new document; and

updating the social network model by modifying relationship strengths in the social network model between the intended entity and the other entities referenced by the new document.

2. The method of claim 1, wherein the contact data is a contacts list.

3. The method of claim 1, wherein the referenced entities comprise people.

4. The method of claim 1, wherein identifying relationships among the referenced entities comprises identifying relationships among the referenced entities responsive to the referenced entities appearing in a same document.

5. The method of claim 1, wherein updating the social networking model comprises modifying a relationship strength between the intended entity and the one of the other identified entities based on a ratio between a number of messages from the intended entity to the one other identified entity and a number of messages from the one other identified entity to the intended entity.

6. The method of claim 1, wherein updating the social networking model comprises modifying a relationship strength between the intended entity and one of the other identified entities based on the frequency with which the intended entity responds to messages from the one other identified entity.

7. The method of claim 1, wherein updating the social networking model comprises modifying a relationship strength between the intended entity and one of the other identified entities based on an elapsed time before the intended entity responds to messages from the one other identified entity.

8. The method of claim 1, wherein the strength of the first relationship is determined from a combination of a directional strength from the first entity to the second entity and a directional strength from the second entity to the first entity.

9. The method of claim 1, wherein performing the name disambiguation comprises:

generating, from the social network model, a list of the at least two candidate entities for the ambiguous entity, the candidate entities being entities from the user'"'"'s social network model;

scoring the candidate entities based at least in part on a strength of a relationship between a candidate entity and the user; and

exposing one or more best scoring candidate entities to the user for selection.

10. The method of claim 9, wherein scoring the candidate entities is also based at least in part on a similarity between words in the new document to words in other documents connected to a candidate entity from the set of documents.

11. The method of claim 9, wherein scoring the candidate entities is also based at least in part on a similarity between the reference to the ambiguous entity and a candidate entity name.

12. The method of claim 1, further comprising performing entity unification between the intended entity and an entity in the user'"'"'s social network model based on the modified relationship strengths.

13. The method of claim 1, further comprising performing entity unification between two entities in the user'"'"'s social network model based on the modified relationship strengths.

14. The method of claim 1, wherein the other identified entities comprise organizations and people, the method further comprising:

adjusting a relationship strength between a person entity and an organization entity based on at least in part on a frequency of mentions of the organization in correspondence to or from the person.

15. The method of claim 1, wherein the other identified entities comprise locations and people, the method further comprising:

adjusting a relationship strength between a person entity and a location entity based at least in part on a frequency of mentions of the location in correspondence to or from the person.

16. A non-transitory computer-readable storage medium storing executable computer program instructions for updating a user'"'"'s social network model, the computer program instructions comprising instructions for:

receiving a set of documents associated with a user;

accessing the user'"'"'s contact data, the contact data identifying a plurality of entities;

analyzing the documents, using the contact data, to identify references to entities therein;

identifying relationships among the referenced entities;

determining a strength of a first relationship between a first entity and a second entity responsive to a volume of documents in which both the first entity and the second entity appear, wherein the first and second entities are a subset of the referenced entities;

building a social network model for the user responsive to the identified relationships among the referenced entities and the strength of the first relationship;

storing the social network model;

receiving a new document associated with the user;

identifying, in the new document, a reference to an ambiguous entity;

performing the name disambiguation using the social network model to determine which of the at least two candidate entities from the social network model is an intended entity for the ambiguous entity;

identifying other entities referenced by the new document; and

updating the social network model by modifying relationship strengths in the social network model between the intended entity and the other entities referenced by the new document.

17. The non-transitory computer-readable storage medium of claim 16, wherein the contact data is a contacts list.

18. The non-transitory computer-readable storage medium of claim 16, wherein the referenced entities comprise people.

19. The non-transitory computer-readable storage medium of claim 16, wherein identifying relationships among the referenced entities comprises identifying relationships among the referenced entities responsive to the referenced entities appearing in a same document.

20. The non-transitory computer-readable storage medium of claim 16, wherein updating the social networking model comprises modifying a relationship strength between the intended entity and one of the other identified entities based on a ratio between a number of messages from the intended entity to the one other identified entity and a number of messages from the one other identified entity to the intended entity.

21. The non-transitory computer-readable storage medium of claim 16, wherein updating the social networking model comprises modifying a relationship strength between the intended entity and one of the other identified entities based on the frequency with which the intended entity responds to messages from the one other identified entity.

22. The non-transitory computer-readable storage medium of claim 16 wherein updating the social networking model comprises modifying a relationship strength between the intended entity and one of the other identified entities based on an elapsed time before the intended entity responds to messages from the one other identified entity.

23. The non-transitory computer-readable storage medium of claim 16, wherein the strength of the first relationship is determined from a combination of a directional strength from the first entity to the second entity and a directional strength from the second entity to the first entity.

24. The non-transitory computer-readable storage medium of claim 16, wherein the computer program instructions further comprise instructions for:

generating, from the social network model, a list of the at least two candidate entities for the ambiguous entity, the candidate entities being entities from the user'"'"'s social network model;

scoring the candidate entities based at least in part on a strength of a relationship between a candidate entity and the user; and

exposing one or more best scoring candidate entities to the user for selection.

25. The non-transitory computer-readable storage medium of claim 24, wherein scoring the candidate entities is also based at least in part on a similarity between words in the new document to words in other documents connected to a candidate entity from the set of documents.

26. The non-transitory computer-readable storage medium of claim 24, wherein scoring the candidate entities is also based at least in part on a similarity between the reference to the ambiguous entity and a candidate entity name.

27. The non-transitory computer-readable storage medium of claim 16, wherein the computer program instructions further comprise instructions for performing entity unification between the intended entity and an entity in the user'"'"'s social network model based on the modified relationship strengths.

28. The non-transitory computer-readable storage medium of claim 16, wherein the computer program instructions further comprise instructions for performing entity unification between two entities in the user'"'"'s social network model based on the modified relationship strengths.

29. The non-transitory computer-readable storage medium of claim 16, wherein the other identified entities comprise organizations and people, and wherein the computer program instructions further comprise instructions for:

adjusting a relationship strength between a person entity and an organization entity based at least in part on a frequency of mentions of the organization in correspondence to or from the person.

30. The non-transitory computer-readable storage medium of claim 16, wherein the other identified entities comprise locations and people, and wherein the computer program instructions further comprise instructions for:

adjusting a relationship strength between a person entity and a location entity based at least in part on a frequency of mentions of the location in correspondence to or from the person.

31. A system for updating a user'"'"'s social network model, the system comprising:

a processor; and

a non-transitory computer-readable storage medium storing computer program instructions executable by the processor, the computer program instructions comprising instructions for;

receiving a set of documents associated with a user;

accessing the user'"'"'s contact data, the contact data identifying a plurality of entities;

analyzing the documents, using the contact data, to identify references to entities therein;

identifying relationships among the referenced entities;

determining a strength of a first relationship between a first entity and a second entity responsive to a volume of documents in which both the first entity and the second entity appear, wherein the first and second entities are a subset of the referenced entities;

building a social network model for the user responsive to the identified relationships among the referenced entities and the strength of the first relationship;

storing the social network model;

receiving a new document associated with the user;

identifying, in the new document, a reference to an ambiguous entity;

performing name disambiguation using the social network model to determine which of at least two candidate entities from the social network model is an intended entity for the ambiguous entity;

identifying other entities referenced by the new document; and

updating the social network model by modifying relationship strengths in the social network model between the intended entity and the other entities referenced by the new document.