Method and system for finding content having a desired similarity

• US 9,195,640 B1
• Filed: 12/23/2011
• Issued: 11/24/2015
• Est. Priority Date: 01/12/2009
• Status: Active Grant

• Current Assignee: SRI International, Inc.
• Original Assignee: SRI International, Inc.
• Inventors: Donneau-Golencer, Thierry D., Hardt, Stephen L.
• Primary Examiner(s): Saint Cyr, Leonard
• Application Number:
  US13/336,360
• Time in Patent Office: 1,432 Days
• Field of Search: 704 7- 10, 704/257
• US Class Current: 1/1
• CPC Class Codes: G06F 16/35 Clustering; Classification
• CPC Class Codes: G06F 40/10 Text processing natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence

Abstract

A method and system for finding content having a desired similarity to an input or source content includes using a similarity model including information and associations derived from content processed by one or more content analyzers to find and/or arrange content having a desired type and/or degree of similarity to the input or source content.

Claims

1. A method, comprising:

maintaining a computerized similarity model comprising information relating to associations of electronic content with other electronic content, the electronic content comprising at least one of textual content and non-textual content, the associations resulting from executing a plurality of computerized content analyzers to extract information from the electronic content, each content analyzer being associated with at least one similarity metric usable to determine a type or degree of similarity of electronic content;

with a computing device, determining a plurality of similarity types based on an input or source content, each similarity type relating to a different characteristic of the input or source content;

weighting each of the plurality of similarity types based on user-specific context information;

identifying, using the similarity model, electronic content that has at least one association in common with the input or source content according to one or more of the similarity types;

ranking the identified electronic content in accordance with the weighting of the similarity types; and

outputting data relating to one or more of the identified electronic content.

2. The method of claim 1, comprising executing a content analyzer configured to identify textual content in the electronic content, identify most important words in the textual content, associate at least one textual element with the electronic content based on the identified most important words, use the at least one textual element to create associations between or among the electronic content, and store the associations between or among the electronic content in the similarity model.

3. The method of claim 1, comprising executing a content analyzer configured to identify textual content in the electronic content, perform semantic analysis of the textual content, associate at least one textual element with the electronic content based on the semantic analysis, use the at least one textual element to create associations between or among the electronic content, and store the associations between or among the electronic content in the similarity model.

4. The method of claim 1, comprising executing a content analyzer configured to determine whether the electronic content comprises at least one Uniform Resource Locator, and in response to determining that the electronic content comprises at least one Uniform Resource Locator, analyzing electronic content associated with the at least one Uniform Resource Locator.

5. The method of claim 1, comprising executing a content analyzer configured to associate the electronic content with at least one topic, associate at least one textual element with the electronic content based on the association of the electronic content with the at least one topic, use the at least one textual element to create associations between or among the electronic content, and store the associations between or among the electronic content in the similarity model.

6. The method of claim 1, comprising executing a content analyzer configured to process electronic content comprising non-textual content, associate at least one textual element comprising textual content relating to the non-textual content with the electronic content, use the at least one textual element to create associations between or among the electronic content, and store the associations between or among the electronic content in the similarity model.

7. The method of claim 6, comprising identifying a graphic in the electronic content, associating at least one textual element comprising textual content relating to the graphic with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

8. The method of claim 6, comprising determining a layout of the electronic content, associating at least one textual element comprising textual content relating to the layout with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

9. The method of claim 6, comprising identifying an image in the electronic content, associating at least one textual element comprising textual content relating to the image with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

10. The method of claim 1, comprising executing a content analyzer configured to associate the electronic content with a style of writing, associating at least one textual element relating to the style of writing with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

11. The method of claim 1, comprising executing a content analyzer configured to identify textual content in the electronic content, associating the textual content with an indicator of sentiment, associating at least one textual element relating to the sentiment with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

12. The method of claim 1, comprising executing a content analyzer configured to associate the electronic content with a subject matter domain, associating at least one textual element comprising textual content relating to the domain with the electronic content, using the at least one textual element to create associations between or among the electronic content, and storing the associations between or among the electronic content in the similarity model.

13. The method of claim 1, comprising executing a content analyzer configured to extract a sequence of textual content from the electronic content, generate a vector representation of the sequence of textual content, the vector representation comprising a plurality of digits, each digit having an association with an aspect of the textual content, associate the vector with the electronic content, use the vector to create associations between or among the electronic content, and store the associations between or among the electronic content in the similarity model.

14. The method of claim 1, comprising determining the desired type or degree of similarity based on at least one of a textual element associated with the input or source content and at least one input from a user.

15. The method of claim 1, comprising monitoring electronic interaction of a user with the electronic content and determining the desired type or degree of similarity based on the user'"'"'s electronic interaction with the electronic content.

16. The method of claim 1, comprising determining the desired type or degree of similarity based on a weight associated with the at least one similarity metric.

17. The method of claim 16, comprising adjusting the weight associated with the at least one of the similarity metric to achieve a lesser or greater degree of similarity.

18. The method of claim 1, comprising selecting at least one content analyzer from the plurality of content analyzers based on the desired type or degree of similarity and executing only the selected at least one content analyzer.

19. The method of claim 1, comprising performing a clustering analysis on the textual elements associated with the electronic content, performing a clustering analysis on textual elements associated with the input or source content, and identifying clusters associated with both the input or source content and at least one of the electronic content.

20. The method of claim 1, comprising ranking electronic content having a common association with the input or source content based on a calculated proximity of the electronic content to the input or source content and interactively arranging the electronic content having a common association with the input or source content based on the ranking.

21. The method of claim 1, wherein the input or source content comprises a plurality of pieces of electronic content, comprising merging the plurality of pieces of electronic content into the input or source content.

22. The method of claim 1, comprising, for each electronic content having an association in common with the input or source content, calculating a proximity score for each of the at least one similarity metric and arranging the electronic content having an association in common with the input or source content based on the proximity score.

23. The method of claim 1, wherein the input or source content is specified by at least one of a user and an automated process.

24. The method of claim 1, comprising associating at least one textual element with the electronic content based on the executing of the plurality of computerized content analyzers.

25. The method of claim 1, comprising executing a first content analyzer to perform a first method of analysis of the electronic content and executing a second content analyzer to perform a second method of analysis of the electronic content.

26. The method of claim 1, comprising maintaining a computerized user-specific profile comprising information relating to at least one of user-specific information and user-specific interactions with electronic content, and determining, using the user-specific profile, whether any of the electronic content has at least one association in common with an input or source content.

27. The method of claim 1, comprising interactively-arranging the electronic content having a common association with the input or source content based on the desired type or degree of similarity of the electronic content to the input or source content.

28. The method of claim 1, comprising interactively-selecting the desired type or degree of similarity used to determine whether the electronic content has a common association with the input or source content.

29. At least one non-transitory computer accessible storage medium comprising a plurality of instructions that in response to being executed, result in a computing device:

analyzing an input or source content using a plurality of content analyzers, each content analyzer associated with at least one similarity metric usable to determine a type or degree of similarity of content,

determining a plurality of similarity types based on the analyzed input or source content, each similarity type relating to a different characteristic of the input or source content;

weighting each of the plurality of similarity types based on user-specific context information;

identifying, based on the analyzing of the input or source content using the plurality of content analyzers, at least one piece of electronic content of a plurality of pieces of electronic content that has an association in common with the input or source content according to one or more of the similarity types;

responsively adjusting the weight associated with the at least one similarity metric based on at least one interaction of a user with the identified electronic content;

ranking the identified electronic content in accordance with the weighting of the similarity types; and

visually arranging the identified electronic content based on the weight associated with the at least one similarity metric.

30. The at least one non-transitory computer accessible storage medium of claim 29, configured to arrange the identified electronic content according to a first similarity metric and re-arrange the identified electronic content according to a second similarity metric.

31. The at least one non-transitory computer accessible storage medium of claim 29, configured to arrange the identified electronic content into a plurality of clusters, select an electronic content from each cluster, and display only the selected electronic content.

32. A computing device comprising:

at least one processor, and

at least one memory device coupled to the at least one processor, the at least one memory device having stored therein a plurality of instructions that when executed by the at least one processor, cause the at least one processor to;

execute a plurality of content analyzers to extract or derive information from each of a plurality of pieces of electronic content determining a plurality of similarity types based on the extracted or derived information, each similarity type relating to a different characteristic of the pieces of electronic content;

weight each of the plurality of similarity types based on user-specific context information;

rank the analyzed electronic content in accordance with the weighting of the plurality of similarity types;

maintain a database comprising information relating to associations of the analyzed electronic content with one or more others of the analyzed electronic content based on results of the executing of the plurality of content analyzers;

receive a search request from a user;

use the database to determine whether any of the analyzed electronic content has a first degree or type of similarity to the search request;

output information relating to at least one possible modification of the first degree or type of similarity;

receive input from the user relating to the at least one possible modification; and

in response to the received input, use the database to determine whether any of the analyzed pieces of content has a second degree or type of similarity to the search request, wherein the second degree or type of similarity relates to a different characteristic of the content than the first degree or type of similarity.

33. The computing device of claim 32, configured to maintain a user-specific database comprising information relating to interactions of a user with electronic content and use the user-specific database to determine whether any of the analyzed pieces of content has the second degree or type of similarity to the search request.