Method and system for retrieving relevant documents from a database

• US 6,370,525 B1
• Filed: 11/13/2000
• Issued: 04/09/2002
• Est. Priority Date: 06/08/1998
• Status: Expired due to Fees

• Current Assignee: Kolomoki Mounds LLC
• Original Assignee: KCSL Incorporated
• Inventors: Kaufman, Ilia
• Primary Examiner(s): Von Buhr, Maria N.
• Application Number:
  US09/711,460
• Time in Patent Office: 512 Days
• Field of Search: 707/1-6
• US Class Current: 707/730
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A method for processing a search query uses the results of a search performed on a high quality, controlled database to assess the relevance of documents retrieved from a search of an uncontrolled public database having documents of highly variable quality. The method includes the steps of parsing the search query and then searching the authoritative database to generate authoritative database results. The search query is also used to search the public database, thereby generating public database results. The quality or relevance of the public database results are then quantified on the basis of the authoritative database results, thereby generating a quality index. The results from both the authoritative and the public databases are then ranked on the basis of this quality index.

Claims

1. A method for ranking a plurality of candidate documents on the basis of the similarity each of the plurality of candidate documents to a user-query, said method comprising the steps of

parsing the user-query, thereby generating a query-word and a distribution of the query-word in the user-query,

assessing an importance of the query-word on the basis of

the frequency with which the query-word occurs in the plurality of candidate documents, and

the distribution of the query-word in the user-query,

evaluating the similarity of a candidate document to the user-query on the basis of a distribution of the query-word in the candidate document, the candidate document having at least one candidate document sentence,

evaluating the similarity of the at least one candidate document sentence to the user-query on the basis of the frequency with which the query-word occurs in the at least one candidate document sentence,

ranking the candidate document relative to the plurality of candidate documents on the basis of the similarity of the candidate document to the user-query and the similarity of the at least one candidate document sentence to the user-query.

2. The method of claim 1 further comprising the step of

evaluating the similarity of the at least-one-candidate-document-sentence to the user-query on the basis of a location of the at-least-one-candidate-document-sentence relative to the public document.

3. The method of claim 1 further comprising the step of generating a derivative query-word, the derivative query-word being selected on the basis of an association with the query-word.

4. The method of claim 1 wherein the step of assessing the importance of the query-word comprises the steps of

determining the overall concentration of the query-word in the plurality of candidate documents,

determining the concentration of the query-word in the user-query,

determining a query-conductance indicative of the importance of the query-word, the query-conductance varying with the concentration of the query-word in the user-query and varying inversely with the overall concentration of the query-word in the plurality of candidate documents.

5. The method of claim 1 wherein the step of evaluating the similarity of the candidate document to the user-query comprises the steps of

determining the concentration of the query-word in the candidate document,

assigning a document-conductance to the candidate document, the document-conductance being selected to vary with the concentration of the query-word in the candidate document.

6. The method of claim 1 wherein the step of evaluating the similarity of the at-least-one-candidate-document-sentence to the user-query comprises the steps of

determining the overall concentration of the query-word in the plurality of candidate documents,

determining the concentration of the query-word in the at-least-one-candidate-document-sentence

assigning a position-independent sentence similarity to the at-least-one-candidate-document-sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the plurality of candidate documents to

the concentration of the query-word in the at-least-one-candidate-document-sentence.

7. A computer-readable medium containing software for ranking a plurality of candidate documents on the basis of the similarity each of the plurality of candidate documents to a user-query, said software comprising instruction for executing the steps of

parsing the user-query, thereby generating a query-word and a distribution of the query-word in the user-query,

assessing an importance of the query-word on the basis of

the frequency with which the query-word occurs in the plurality of candidate documents, and

the distribution of the query-word in the user-query,

evaluating the similarity of a candidate document to the user-query on the basis of a distribution of the query-word in the candidate document, the candidate document having at least one candidate document sentence,

evaluating the similarity of the at least one candidate document sentence to the user-query on the basis of the frequency with which the query-word occurs in the at least one candidate document sentence,

ranking the candidate document relative to the plurality of candidate documents on the basis of the similarity of the candidate document to the user-query and the similarity of the at least one candidate document sentence to the user-query.

8. The computer readable medium of claim 7 wherein the software further comprises instructions for executing the step of

evaluating the similarity of the at-least-one-candidate-document-sentence to the user-query on the basis of a location of the at-least-one-candidate-document-sentence relative to the public document.

9. The computer readable medium of claim 7 wherein the software further comprises instructions for executing the step of generating a derivative query-word, the derivative query-word being selected on the basis of an association with the query-word.

10. The computer readable medium of claim 7 wherein the software further comprises instructions for executing the step of

assessing an importance of the derivative query-word on the basis of

the frequency with which the derivative query-word occurs in the plurality of candidate documents, and

the distribution of the derivative query-word in the user-query.

11. The computer readable medium of claim 10 wherein the software further comprises instructions for executing the step of diminishing the importance of the derivative query-word relative to the importance of the query-word.

12. The computer readable medium of claim 7 wherein the software for executing the step of assessing the importance of the query-word comprises instructions for executing the steps of

determining the overall concentration of the query-word in the plurality of candidate documents,

determining the concentration of the query-word in the user-query,

determining a query-conductance indicative of the importance of the query-word, the query-conductance varying with the concentration of the query-word in the user-query and varying inversely with the overall concentration of the query-word in the plurality of candidate documents.

13. The computer readable medium of claim 7 wherein the software for executing the step of evaluating the similarity of the candidate document to the user-query comprises software instructions for executing the steps of

determining the concentration of the query-word in the candidate document,

assigning a document-conductance to the candidate document, the document-conductance being selected to vary with the concentration of the query-word in the candidate document.

14. The computer readable medium of claim 7 wherein the software for executing the step of evaluating the similarity of the at-least-one-candidate-document-sentence to the user-query comprises software instruction for executing the steps of

determining the overall concentration of the query-word in the plurality of candidate documents,

determining the concentration of the query-word in the at-least-one-candidate-document-sentence

assigning a position-independent sentence similarity to the at-least-one-candidate-document-sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the plurality of candidate documents to

the concentration of the query-word in the at-least-one-candidate-document-sentence.

15. The computer-readable medium of claim 7 wherein the software further comprises instructions for executing the step of retrieving the at-least-one authoritative document from a local storage device.

16. The computer-readable medium of claim 7 wherein the software further comprises instructions for executing the step of retrieving the at-least one authoritative document from a server on a network.

17. The computer-readable medium of claim 16 wherein the software further comprises instructions for selecting the network to be a corporate-wide network.

18. The computer-readable medium of claim 16 wherein the software further comprises instructions for selecting the network to be a global computer network.

19. The computer-readable medium claim 7 wherein the software further comprises instructions for executing the step of selecting the authoritative database from a plurality of authoritative databases.

20. A computer-implemented method for generating a list of candidate documents, said method comprising the steps of

evaluating, on the basis of a search query, a relevance of an authoritative document selected from an authoritative database,

evaluating, on the basis of the search query, a relevance of a public document from a public database, and

including the public document in the list of candidate documents if the relevance of the public document exceeds a relevance threshold selected on the basis of the relevance of the authoritative document.

21. The method of claim 20 wherein the first evaluating step comprises the steps of

parsing the search query, thereby retrieving a query word and a distribution of the query word in the user-query, and

assessing an importance of the query word on the basis of the authoritative document.

22. The method of claim 21 wherein the assessing step comprises the step of assessing the importance of the query word on the basis of the distribution of query words in the search query and the frequency with which the query word occurs in the public document.

23. The method of claim 21 wherein the step of evaluating a relevance of a public document comprises the step of evaluating the relevance of the public document on the basis of the distribution of the query word in the public document.

24. The method of claim 23 wherein the public document includes at least one sentence and the step of evaluating the relevance of the public document on the basis of the distribution of the query word comprises the step of evaluating the frequency with which the query word occurs in the at least one sentence.

25. The method of claim 21 wherein the step of assessing the importance of the query word comprises the steps of

determining an overall concentration in the authoritative database of the query-word,

determining a concentration of the query-word in the search-query, and

determining a query-conductance indicative of the importance of the query-word, the query-conductance varying with the concentration of the query-word in the search-query and varying inversely with the overall concentration of the query-word in the authoritative database.

26. The method of claim 20 wherein the step of evaluating the relevance of the authoritative document to the search-query comprises the steps of

determining the concentration in the authoritative document of the query-word,

assigning a document-conductance to the authoritative document, the document-conductance being selected to vary with the concentration of the query-word in the authoritative document.

27. The method of claim 20 wherein the step of evaluating the similarity of the public document to the search-query comprises the steps of

determining the concentration of the query-word in the public document,

assigning a document-conductance to the public document, the document-conductance being selected to vary with the concentration of the query-word in the public document.

28. The method of claim 20 wherein the authoritative document includes at-least-one-authoritative-document sentence and the step of evaluating the relevance of the authoritative document to the search-query comprises the steps of

determining the overall concentration of the query-word in the authoritative database,

determining the concentration of the query-word in the at-least-one-authoritative-document sentence

assigning a position-independent sentence similarity to the at-least-one-authoritative-document sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the authoritative database to

the concentration of the query-word in the at-least-one-authoritative-document sentence.

29. The method of claim 20 wherein the public document includes at-least-one-public-document sentence and the step of evaluating the relevance of a public document to the user-query comprises the steps of

determining the overall concentration of the query-word in the authoritative database,

determining the concentration of the query-word in the at-least-one-public-document sentence

assigning a position-independent sentence similarity to the at-least-one-public-document sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the public database to

the concentration of the query-word in the at-least-one-public-document-sentence.

30. The method of claim 20 further comprising the step of retrieving the authoritative document from a local storage device.

31. The method of claim 20 further comprising the step of retrieving the authoritative document from a network.

32. The method of claim 31 further comprising the step of selecting the network to be a corporate-wide network.

33. The method of claim 31 further comprising the step of selecting the network to be a global computer network.

34. A computer-readable medium containing software for generating a list of candidate documents, the software comprising instructions for executing the steps of

evaluating, on the basis of a search query, a relevance of an authoritative document selected from an authoritative database,

evaluating, on the basis of the search query, a relevance of a public document, and

including the public document in the list of candidate documents if the relevance of the public document exceeds a relevance threshold selected on the basis of the relevance of the authoritative document.

35. The computer-readable medium of claim 34 wherein the instructions for executing the first evaluating step comprise instructions for executing the steps of

parsing the search query, thereby retrieving a query word and a distribution of the query word in the user-query, and

assessing an importance of the query word on the basis of the authoritative document.

36. The computer-readable medium of claim 35 wherein the instructions for executing the assessing step comprise instructions for executing the step of assessing the importance of the query word on the basis of the distribution of query words in the search query and the frequency with which the query word occurs in the public document.

37. The computer-readable medium of claim 35 wherein the instructions for executing the step of evaluating a relevance of a public document comprise instructions for executing the step of evaluating the relevance of the public document on the basis of the distribution of the query word in the public document.

38. The computer-readable medium of claim 37 wherein the public document includes at least one sentence and the instructions for executing the step of evaluating the relevance of the public document on the basis of the distribution of the query word comprise instructions for executing the step of evaluating the frequency with which the query word occurs in the at least one sentence.

39. The computer-readable medium of claim 35 wherein the instructions for executing the step of assessing the importance of the query word comprise instructions for executing the steps of

determining an overall concentration in the authoritative database of the query-word,

determining a concentration of the query-word in the search-query, and

determining a query-conductance indicative of the importance of the query-word, the query-conductance varying with the concentration of the query-word in the search-query and varying inversely with the overall concentration of the query-word in the authoritative database.

40. The computer-readable medium of claim 34 wherein the instructions for executing the step of evaluating the relevance of the authoritative document to the search-query comprise instructions for executing the steps of

determining the concentration in the authoritative document of the query-word,

assigning a document-conductance to the authoritative document, the document-conductance being selected to vary with the concentration of the query-word in the authoritative document.

41. The computer-readable medium of claim 34 wherein the instructions for executing the step of evaluating the similarity of the public document to the search-query comprise instructions for executing the steps of

determining the concentration of the query-word in the public document,

assigning a document-conductance to the public document, the document-conductance being selected to vary with the concentration of the query-word in the public document.

42. The computer-readable medium of claim 34 wherein the authoritative document includes at least one authoritative-document sentence and said instructions for executing the step of evaluating the relevance of the authoritative document to the search-query comprise instructions for executing the steps of

determining the overall concentration of the query-word in the authoritative database,

determining the concentration of the query-word in the at-least-one-authoritative-document sentence

assigning a position-independent sentence similarity to the at-least-one-authoritative-document sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the authoritative database to

the concentration of the query-word in the at-least-one-authoritative-document sentence.

43. The computer-readable medium of claim 34 wherein the public document includes at least one public-document sentence and the instructions for executing the step of evaluating the relevance of a public document to the user-query comprise instructions for executing the steps of

determining the overall concentration of the query-word in the authoritative database,

determining the concentration of the query-word in the at-least-one-public-document sentence

assigning a position-independent sentence similarity to the at-least-one-public-document sentence, the position-independent sentence similarity being responsive to the ratio of

the overall concentration of the query-word in the public database to

the concentration of the query-word in the at-least-one-public-document-sentence.

44. The computer-readable medium of claim 34 further comprising instructions for executing the step of retrieving the authoritative document from a local storage device.

45. The computer-readable medium of claim 34 further comprising instructions for executing the step of retrieving the authoritative document from a network.

46. The computer-readable medium of claim 45 further comprising instructions for executing the step of selecting the network to be a corporate-wide network.

47. The computer-readable medium of claim 45 further comprising instructions for executing the step of selecting the network to be a global computer network.