Multiple engine information retrieval and visualization system

• US 6,701,318 B2
• Filed: 02/03/2003
• Issued: 03/02/2004
• Est. Priority Date: 11/18/1998
• Status: Expired due to Fees

• Current Assignee: Technology Licensing Corporation
• Original Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Inventors: Cusick, Gregory J., Snowberg, Eric J., Fox, Kevin L., Frieder, Ophir, Knepper, Margaret M., Killam, Robert A., Nemethy, Joseph M.
• Primary Examiner(s): ALAM, SHAHID AL
• Application Number:
  US10/356,958
• Publication Number:
  US 20030130998A1
• Time in Patent Office: 393 Days
• Field of Search: 707/5, 707/10, 707/100, 707/102, 705/35-38, 345/764, 709/217, 709/218, 709/219, 715/513
• US Class Current: 707/706
• CPC Class Codes: G06F 16/338 Presentation of query results
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: Y10S 707/917 Text
• CPC Class Codes: Y10S 707/959 Network
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface

Abstract

An information retrieval and visualization system utilizes multiple search engines for retrieving documents from a document database based upon user input queries. Search engines include an n-gram search engine and a vector space model search engine using a neural network training algorithm. Each search engine produces a common mathematical representation of each retrieved document. The retrieved documents are then combined and ranked. Mathematical representations for each respective document is mapped onto a display. Information displayed includes a three-dimensional display of keywords from the user input query. The three-dimensional visualization capability based upon the mathematical representation of information within the information retrieval and visualization system provides users with an intuitive understanding, with relevance feedback/query refinement techniques that can be better utilized, resulting in higher retrieval accuracy (precision).

Claims

1. An information retrieval system for selectively retrieving documents from a document database, the system comprising:

an input interface for accepting at least one user search query;

an n-gram search engine for retrieving documents from the document database based upon the at least one user search query, said n-gram search engine producing a common mathematical representation of each retrieved document;

a vector space model (VSM) search engine for retrieving documents from the document database based upon the at least one user search query, said VSM search engine producing a common mathematical representation of each retrieved document;

a display; and

visualization display means for mapping respective mathematical representations of the retrieved documents onto said display.

2. An information retrieval system according to claim 1, wherein said VSM search engine produces a document context vector representation and an axis context vector representation of each retrieved document;

and wherein said visualization display means maps the axis context vector representations and the document context vector representations of the retrieved documents onto said display.

3. An information retrieval system according to claim 1, wherein the document context vector is based upon a sum of all words in a document after reducing low content words.

4. An information retrieval system according to claim 1, wherein the axis context vector is based upon a sum of all words in each axis after reducing low content words.

5. An information retrieval system according to claim 1, wherein said visualization display means further comprises keyword display means for providing a three-dimensional display of keywords from the at least one user input query.

6. An information retrieval system according to claim 5, wherein said visualization display means further comprises cluster display means for displaying retrieved documents in clusters surrounding respective user search queries.

7. An information retrieval system according to claim 6, wherein the retrieved documents are displayed in a color different from a color used for displaying the keywords from the at least one user search query.

8. An information retrieval system according to claim 6, wherein said visualization display means further comprises means for viewing different aspects of the retrieved documents on the three-dimensional display.

9. An information retrieval system according to claim 1, wherein said visualization display means provides a list of retrieved documents, and wherein each retrieved document has an assigned score indicating relevance to the search query with respect to the other retrieved documents.

10. An information retrieval system according to claim 1, wherein a retrieved document mapped onto said display is selectable via said input interface, with the text of the selected document being displayed on said display.

11. An information retrieval system according to claim 1, further comprising fusing means for combining and ranking retrieved documents from said n-gram search engine and said VSM search engine.

12. An information retrieval system according to claim 1, wherein said n-gram search engine comprises n-gram training means for least frequency training of training documents.

13. An information retrieval system according to claim 1, wherein said VSM search engine comprises VSM training means for processing training documents;

and a neural network.

14. An information retrieval system according to claim 1, wherein each search engine comprises a plurality of user selectable dictionaries for determining mathematical representations of documents.

15. An information retrieval system according to claim 1, wherein said input interface comprises relevance feedback means for accepting relevance feedback from the user.

16. An information retrieval system according to claim 15, wherein said relevance feedback means comprises means for selecting one or more retrieved documents as a next search query.

17. An information retrieval system according to claim 1, wherein said input interface comprises means for permitting a user to assign a weighting percentage to each search engine.

18. An information retrieval system according to claim 1, wherein the at least one user search query comprises at least one keyword.

19. An information retrieval system according to claim 1, wherein the at least one user search query comprises at least one document.

20. An information retrieval system according to claim 1, wherein the at least one user search query comprises at least one document cluster.

21. A method for selectively retrieving documents from a document database using an information retrieval system comprising an n-gram search engine and a vector space model (VSM) search engine, the method comprising:

generating at least one user search query;

retrieving documents from the document database using the n-gram search engine based upon the at least one user search query, the n-gram search engine producing a common mathematical representation of each retrieved document;

retrieving documents from the document database using the VSM search engine based upon the at least one user search query, the VSM search engine producing a common mathematical representation of each retrieved document; and

mapping respective mathematical representations of the retrieved documents onto a display.

22. A method according to claim 21, wherein retrieving documents from the document database using the VSM search engine comprises:

producing a document context vector representation of each retrieved document; and

producing an axis context vector representation of each retrieved document.

23. A method according to claim 22, wherein the mapping comprises mapping the axis and the document context vector representations of the retrieved documents onto the display.

24. A method according to claim 22, wherein the document context vector is based upon a sum of all words in a document after reducing low content words.

25. A method according to claim 22, wherein the axis context vector is based upon a sum of all words in each axis after reducing low content words.

26. A method according to claim 21, wherein the mapping further comprising displaying keywords from the at least one user search query onto a three dimensional display.

27. A method according to claim 26, wherein the mapping comprises displaying retrieved documents in clusters surrounding respective user search queries.

28. A method according to claim 27, wherein the displaying comprises displaying retrieved documents in a color different from a color used for displaying the keywords from the at least one user search query.

29. A method according to claim 27, wherein the displaying comprises displaying different aspects of the retrieved documents on the three-dimensional display.

30. A method according to claim 21, further comprising providing a list of retrieved documents, and wherein each retrieved document has an assigned score indicating relevance to the search query with respect to the other retrieved documents.

31. A method according to claim 21, further comprising:

receiving an input for selecting a retrieved document mapped onto the display; and

displaying the text of the selected document on the display.

32. A method according to claim 21, further comprising combining and ranking retrieved documents from the n-gram search engine and the VSM search engine.

33. A method according to claim 21, wherein the n-gram search engine comprises n-gram training means for least frequency training of training documents.

34. A method according to claim 21, wherein the VSM search engine comprises VSM training means for processing training documents.

35. A method according to claim 21, further comprising providing relevance feedback means for accepting relevance feedback from a user.

36. A method according to claim 35, wherein providing relevance feedback means comprises selecting one or more retrieved documents as a next search query.

37. A method according to claim 21, further comprising assigning a weighting percentage to each search engine.

38. A method according to claim 21, wherein the generating comprises generating at least one keyword.

39. A method according to claim 21, wherein the generating comprises generating at least one document.

40. A method according to claim 21, wherein the generating comprises generating at least one document cluster.