Recommending search terms using collaborative filtering and web spidering

• US 20030055816A1
• Filed: 12/11/2001
• Published: 03/20/2003
• Est. Priority Date: 05/28/1999
• Status: Active Grant

• Current Assignee: R2 Solutions LLC (Acacia Research Corporation)
• Original Assignee: Overture Services Incorporated (Apollo Global Management, Inc.)
• Inventors: Davis, Darren, Geddis, Don F., Davies, Winton, Paine, Mark, Dukes-Schlossberg, Jon
• Granted Patent: US 7,225,182 B2
• Time in Patent Office: Days
• US Class Current: 707/3
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/36 Creation of semantic tools, e.g. ontology or thesauri
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: Y10S 707/918 Location
• CPC Class Codes: Y10S 707/959 Network
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching

Abstract

In a pay-for-placement search system, the system makes search term recommendations to advertisers managing their accounts in one or more of two ways. A first technique involves looking for good search terms directly on an advertiser'"'"'s web site. A second technique involves comparing an advertiser to other, similar advertisers and recommending the search terms the other advertisers have chosen. The first technique is called spidering and the second technique is called collaborative filtering. In the preferred embodiment, the output of the spidering step is used as input to the collaborative filtering step. The final output of search terms from both steps is then interleaved in a natural way.

Claims

1. A method for a database search system, the method comprising:

receiving a list of search terms associated with an advertiser on the database search system;

determining candidate search terms based on search terms of other advertisers on the database search system; and

recommending the additional search terms from among the candidate search terms.

2. The method of claim 1 wherein determining candidate search terms comprises:

assigning ratings to search terms; and

computing a correlation between the advertiser and one or more of the other advertisers using the assigned ratings of advertiser search terms.

3. The method of claim 2 further comprising:

predicting a likelihood that a candidate search term will be relevant to the advertiser.

4. The method of claim 3 wherein predicting comprises:

determining a quality metric for the candidate search terms; and

predicting relevance of candidate search terms based on the quality metric.

5. A method for a database search system, the method comprising:

maintaining a database of search listings including associated search terms;

receiving a list of search terms associated with an advertiser;

computing ratings for search terms; and

recommending additional search terms to the advertiser based on the computed ratings.

6. The method of claim 5 wherein computing ratings comprises:

assigning ratings to search terms.

7. The method of claim 5 wherein computing ratings comprises:

predicting ratings for search terms.

8. The method of claim 5 wherein receiving a list of search terms comprises:

receiving a list of initial search terms from the advertiser.

9. The method of claim 5 wherein receiving a list of search terms comprises:

identifying an existing advertiser on the database search system; and

forming the list of search terms from search terms of the existing advertiser.

10. The method of claim 5 wherein receiving a list of search terms comprises:

receiving a website uniform resource locator (URL); and

spidering the website associated with the URL to extract search terms for the list of search terms.

11. The method of claim 10 wherein spidering the website comprises:

receiving data from pages of the website;

recording candidate search terms from the data; and

determining a quality metric for each candidate search term.

12. The method of claim 11 further comprising:

sorting the candidate search terms according to the quality metric; and

recommending only candidate search terms having a quality metric exceeding a threshold.

13. The method of claim 10 wherein spidering comprises:

receiving data from one or more pages of the website; and

examining text from the one or more pages for candidate search terms.

14. The method of claim 13 wherein examining text comprises:

examining substantially all text from the one or more pages; and

examining meta tags from the one or more pages.

15. The method of claim 13 wherein receiving a website URL comprises:

receiving the advertiser'"'"'s URL as the web site URL.

16. The method of claim 13 wherein receiving a website URL comprises:

receiving the web site URL from the advertiser.

17. The method of claim 5 further comprising:

receiving a search query from a user;

searching for matching search terms in the database of search listings;

preparing search results by formatting matching search terms according to advertiser bid amounts associated with the search listings; and

communicating the search results to the user.

28. The database search system of claim 15 further comprising:

search engine program code configured to search the database in response to a search query from a user.

18. A database search system comprising:

a database of search terms, each search term being associated with one or more advertisers; and

program code configured to recommend additional search terms for an advertiser based on search terms in the database.

19. The database search system of claim 18 wherein the program code comprises:

collaborative filtering code configured to recommend the additional search terms based on search terms associated with other advertisers of the database search system.

20. The database search system of claim 19 wherein the collaborative filtering code comprises Pearson correlation code.

21. The database search system of claim 19 wherein the program code comprises:

a program loop to iterate the collaborative filtering code.

22. The database search system of claim 18 further comprising:

code to receive accept and reject indications from the advertiser before repeating the program loop.

23. The database search system of claim 18 wherein the program code comprises:

spidering code to recommend the additional search terms.

24. The database search system of claim 18 wherein the program code comprises:

spidering code to find initially accepted search terms in a web site; and

collaborative filtering code to provide the recommended additional search terms.

25. The database search system of claim 24 wherein the spidering code is configured to spider a web site of the advertiser.

26. The database search system of claim 23 wherein the spidering code is configured to spider a web site specified by the advertiser.

27. The database search system of claim 18 further comprising:

filtering code to filter candidate search terms according to a quality metric to produce the recommended additional search terms.

29. A data base operating method for a database search system storing advertiser search listing, the method comprising:

spidering a specified web site to obtain an initial list of advertiser search terms for an advertiser;

filtering the initial list of advertiser search terms using search terms of other advertisers; and

storing in a search listing database search listings for the advertiser, the search listings formed with the filtered search terms.

30. The data base operating method of claim 29 wherein the specified web site comprises an advertiser specified website.

31. The data base operating method of claim 29 wherein the specified web site comprises a website specified by the database search system.

32. The data base operating method of claim 29 further comprising:

assigning ratings to search terms; and

computing a correlation between the advertiser and one or more of the other advertisers using the assigned ratings of advertiser search terms.

33. The data base operating method of claim 32 further comprising:

.

predicting a likelihood that a search term will be relevant to the advertiser.

34. The data base operating method of claim 33 wherein predicting comprises:

determining a quality metric for candidate search terms; and

predicting relevance of candidate search terms based on the quality metric.

35. The data base operating method of claim 29 wherein spidering the specified web site comprises:

receiving data from pages of the specified website;

recording candidate search terms from the data; and

determining a quality metric for each candidate search term.

36. The data base operating method of claim 35 further comprising:

sorting the candidate search terms according to the quality metric; and

recommending only candidate search terms having a quality metric exceeding a threshold.

37. The data base operating method of claim 29 wherein filtering comprises:

determining a correlation between a web site of the advertiser and web sites of other advertisers on the database search system;

using the correlation, determining ratings for each advertiser search term in the initial list of advertiser search terms; and

organizing search terms of the initial list of advertiser search terms according to the ratings.

38. The data base operating method of claim 29 further comprising:

presenting the organized search terms to the advertiser;

receiving advertiser acceptance indications for the organized search terms;

adjusting the list of advertiser search terms according to the acceptance indications; and

filtering the adjusted list.

39. The data base operating method of claim 38 further comprising:

iteratively presenting the organized search terms, receiving revised advertiser acceptance indications, and adjusting the list of advertiser search terms.

40. The data base operating method of claim 38 further comprising:

receiving a search query from a user;

searching for matching search terms in the search listing database;

preparing search results by formatting matching search terms according to advertiser bid amounts associated with the matching search listings; and

communicating the search results to the user.

43. The method of claim 38 wherein generating the list of additional search terms comprises:

matching one or more text strings from the received search term with a thesaurus.

41. A method for a database search system, the method comprising:

receiving a search term of an advertiser;

in response to the received search term, generating a list of additional related search terms; and

receiving advertiser selected search terms from the list of additional related search terms.

42. The method of claim 41 wherein generating the list of additional search terms comprises:

matching one or more text strings from the received search term with a database of search terms.

44. The method of claim 41 further comprising:

displaying a form for entering one or more search listings for a selected search term.

45. The method of claim 42 further comprising:

entering the selected search term as a default value in each of the one or more search listings

46. A method for recommending search terms, the method comprising:

receiving information describing an advertiser; and

recommending search terms based on the advertiser'"'"'s similarity to other advertisers.

47. The method of claim 46 further comprising:

assigning ratings based on the information describing the advertiser and information describing the other advertisers; and

recommending search terms based at least in part on the ratings.

48. The method of claim 46 further comprising:

predicting ratings for the advertiser and the other advertisers based on the information describing the advertiser and information describing the other advertisers; and

recommending search terms based at least in part on the ratings.

49. The method of claim 46 further comprising:

computing correlations for the advertiser and the other advertisers based on the information describing the advertiser and information describing the other advertisers; and

recommending search terms based at least in part on the correlations.

50. The method of claim 49 wherein computing correlations comprises:

computing Pearson correlations.

51. A method for recommending search terms, the method comprising:

downloading web pages rooted at a specified uniform resource locator (URL); and

recommending search terms that appear in the body text of the web pages.

52. The method of claim 51 further comprising:

identifying an advertiser having a web site;

specifying the advertiser'"'"'s web site as the specified URL.

53. The method of claim 51 further comprising:

receiving a web site identifier from an advertiser as the specified URL.

54. A method for recommending search terms, the method comprising:

receiving a uniform resource locator (URL);

downloading web pages rooted at the URL; and

recommending search terms that appear in the web pages.

55. The method of claim 54 further comprising:

identifying an advertiser having a web site;

receiving the advertiser'"'"'s web site as the URL.

56. A method for recommending search terms, the method comprising:

calculating a quality metric for candidate search terms, the quality metric for a respective candidate search term being a function of the respective search term'"'"'s web frequency and a function of a search term'"'"'s search frequency; and

recommending search terms for which the calculated quality metric exceeds a quality threshold

57. The method of claim 56 further comprising:

sorting the candidate search terms according to a second metric.

58. The method of claim 56 further comprising:

automatically calibrating the quality threshold.

59. A method for recommending search terms for a database search system, the method comprising:

maintaining a database of search terms associated with advertiser listings;

receiving information describing an advertiser; and

recommending search terms to the advertiser.

60. The method of claim 59 further comprising:

determining candidate search terms by collaborative filtering; and

recommending search terms from the candidate search terms.

61. The method of claim 59 further comprising:

downloading web pages rooted at a uniform resource locator; and

recommending the search terms based on terms that appear in body text of the web pages.

62. The method of claim 59 further comprising:

receiving a uniform resource locator (URL);

downloading web pages rooted at the URL; and

recommending search terms based on terms that appear in the web pages.

63. The method of claim 59 further comprising:

calculating a quality metric for candidate search terms, the quality metric for a respective candidate search term being a function of the respective search term'"'"'s web frequency and a function of a search term'"'"'s search frequency; and

recommending the search terms based on search terms for which the calculated quality metric exceeds a quality threshold

65. The method of claim 59 further comprising:

receiving feedback from the advertiser on the recommended search terms; and

changing the recommended search terms based on the feedback.

64. The method of claim 59 further comprising:

receiving feedback from the advertiser on the recommended search terms; and

changing the recommended search terms based on the feedback.