Multiple engine information retrieval and visualization system

• US 20030130998A1
• Filed: 02/03/2003
• Published: 07/10/2003
• Est. Priority Date: 11/18/1998
• Status: Active Grant

• Current Assignee: Technology Licensing Corporation
• Original Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Inventors: Frieder, Ophir, Knepper, Margaret M., Fox, Kevin L., Nemethy, Joseph M., Killam, Robert A., Cusick, Gregory J., Snowberg, Eric J.
• Granted Patent: US 6,701,318 B2
• Time in Patent Office: Days
• US Class Current: 707/3
• 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;

a plurality of search engines for retrieving documents from the document database based upon the at least one user search query, each of said search engines 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 at least one of said search engines produces a document context vector representation and an axis context vector representation of each retrieved document;

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

3. An information retrieval system according to claim 2, 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 2, 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 search 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 all of said search engines.

12. An information retrieval system according to claim 1, wherein said plurality of search engine comprises an n-gram search engine.

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

14. An information retrieval system according to claim 1, wherein said plurality of search engines comprises a vector space model (VSM) search engine.

15. An information retrieval system according to claim 14, wherein said VSM search engine comprises VSM training means for processing training documents, and comprising a neural network.

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

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

18. An information retrieval system according to claim 17, wherein said relevance feedback means comprises means for selecting one or more retrieved documents as a next search query.

19. 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.

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

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

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

23. 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;

a 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.

24. An information retrieval system according to claim 23, 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 means maps the axis context vector representations of the retrieved documents onto said display.

25. An information retrieval system according to claim 23, wherein the document context vector is based upon a sum of all words in a document after reducing low content words.

26. An information retrieval system according to claim 23, wherein the axis context vector is based upon a sum of all words in each axis after reducing low content words.

27. An information retrieval system according to claim 23, 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.

28. An information retrieval system according to claim 27, wherein said visualization display means further comprises cluster display means for displaying retrieved documents in clusters surrounding respective user search queries.

29. An information retrieval system according to claim 28, 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.

30. An information retrieval system according to claim 28, wherein said visualization display means further comprises means for viewing different aspects of the retrieved documents on the three-dimensional display.

31. An information retrieval system according to claim 23, 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.

32. An information retrieval system according to claim 23, 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.

33. An information retrieval system according to claim 23, further comprising fusing means for combining and ranking retrieved documents from said n-gram search engine and said VSM search engine.

34. An information retrieval system according to claim 23, wherein said n-gram search engine comprises n-gram training means for least frequency training of training documents.

35. An information retrieval system according to claim 23, wherein said VSM search engine comprises VSM training means for processing training documents, and comprising a neural network.

36. An information retrieval system according to claim 23, wherein each search engine comprises a plurality of user selectable dictionaries for determining mathematical representations of documents.

37. An information retrieval system according to claim 23, wherein said input interface comprises relevance feedback means for accepting relevance feedback from the user.

38. An information retrieval system according to claim 37, wherein said relevance feedback means comprises means for selecting one or more retrieved documents as a next search query.

39. An information retrieval system according to claim 23, wherein said input interface comprises means for permitting a user to assign a weighting percentage to each search engine.

40. An information retrieval system according to claim 23, wherein the at least one user search query comprises at least one keyword.

41. An information retrieval system according to claim 23, wherein the at least one user search query comprises at least one document.

42. An information retrieval system according to claim 23, wherein the at least one user search query comprises at least one document cluster.

43. A method for selectively retrieving documents from a document database using an information retrieval system comprising a plurality of search engines, the method comprising the steps of:

generating at least one user search query;

retrieving documents from the document database based upon the user search query, with each search engine searching the document database;

producing a common mathematical representation of each document retrieved by the respective search engines; and

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

44. A method according to claim 43, wherein the step of producing comprises the steps of:

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

producing an axis context vector representation of each retrieved document.

45. A method according to claim 44, wherein the document context vector is based upon a sum of all words in a document after reducing low content words.

46. A method according to claim 44, wherein the axis context vector is based upon a sum of all words in each axis after reducing low content words.

47. A method according to claim 44, wherein the step of mapping comprises the step of mapping axis the context vector representations of the retrieved documents onto the display.

48. A method according to claim 44, wherein the step of mapping comprises the step of comparing the document context vectors with the at least one user search query.

49. A method according to claim 43, wherein the step of mapping further comprising the step of displaying keywords from the at least one user input query onto a three dimensional display.

50. A method according to claim 49, wherein the step of mapping comprises the step of displaying retrieved documents in clusters surrounding respective user search queries.

51. A method according to claim 50, wherein the step of displaying comprises the step of displaying retrieved documents in a color different from a color used for displaying the keywords from the at least one user search query.

52. A method according to claim 50, wherein the step of displaying comprises the step of displaying different aspects of the retrieved documents on the three-dimensional display.

53. A method according to claim 43, further comprising the step of 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.

54. A method according to claim 43, further comprising the steps of:

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

displaying the text of the selected document on the display.

55. A method according to claim 43, further comprising the step of combining and ranking retrieved documents from all of the search engines.

56. A method according to claim 43,wherein the plurality of search engines comprises an n-gram search engine, and further comprising the step of least frequency training of training documents.

57. A method according to claim 43, wherein the plurality of search engines comprises a vector space model (VSM) search engine, and further comprising the step of processing training documents.

58. A method according to claim 43 further comprising the step of providing relevance feedback means for accepting relevance feedback from a user.

59. A method according to claim 58, wherein the step of providing relevance feedback means comprises the step of selecting one or more retrieved documents as a next search query.

60. A method according to claim 43, further comprising the step of assigning a weighting percentage to each search engine.

61. A method according to claim 43, wherein the step of generating comprises the step of generating at least one keyword.

62. A method according to claim 43, wherein the step of generating comprises the step of generating at least one document.

63. A method according to claim 43, wherein the step of generating comprises the step of generating at least one document cluster.

64. A method for selectively retrieving documents from a document database, the method comprising the steps of:

defining a dictionary comprising a plurality of words related to a topic to be searched;

randomly assigning a context vector to each word in the dictionary;

training the dictionary words;

assigning axis representation to each dictionary word;

receiving at least one user search query; and

searching a document database based upon the at least one user search query.

65. A method according to claim 64, wherein the step of training comprises the steps of:

creating context vectors for each word in at least one training document; and

converging the context vectors toward each other for the context vectors representing words appearing close to one another based upon contextual usage.

66. A method according to claim 64, wherein the step of assigning comprises the step of assigning each dictionary word to an axis having a largest component of a respective context vector.

67. A method according to claim 64, further comprising the step of displaying a mathematical representation of the retrieved documents from the document database corresponding to the at least one user search query.

68. A method according to claim 67, wherein the step of displaying comprises displaying three-dimensional keywords from the at least one user search query in three-dimensions.

69. A method according to claim 68, wherein the step of displaying comprises displaying retrieved documents in clusters surrounding respective keywords.

70. A method according to claim 69, wherein the step of displaying comprises displaying retrieved documents in a color different from a color used for displaying keywords from the at least one user search query.

71. A method according to claim 67, wherein the step of displaying comprises displaying different aspects of the retrieved documents on the three-dimensional display.

72. A method according to claim 64, further comprising the step of generating a list of retrieved documents, each having an assigned score indicating relevance to the at least one user search query with respect to other retrieved documents.

73. A method according to claim 67, further comprising the steps of:

receiving an input for selecting a retrieved document; and

displaying the text of the selected document.

74. A method according to claim 67, where the step of displaying comprises the step of multiplying an axis representation of the search query by an axis representation of a retrieved document.

75. A method according to claim 64, where the step of assigning comprises the steps of:

calculating a cosine angle between the axis representation of each dictionary word with each context vector; and

assigning each dictionary word to a respective context vector corresponding to an axis having the highest calculated cosine angle.

76. A method according to claim 64, further comprising the step of receiving relevance feedback from a user.

77. A method according to claim 76, further comprising the step of selecting one or more retrieved documents as a next search query.

78. A method according to claim 64, wherein the step of searching comprises the step of using an n-gram search engine for searching the document database.

79. A method according to claim 64, wherein the step of searching comprises the step of using a vector space model (VSM) for searching the document database.