Search boost vector based on co-visitation information

• US 8,874,570 B1
• Filed: 11/30/2004
• Issued: 10/28/2014
• Est. Priority Date: 11/30/2004
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Macduff, Tessa, Haveliwala, Taher H., Jeh, Glen, Kamvar, Sepandar D., Oztekin, Bilgehan Uygar, Sahami, Mehran, Basu, Sugato
• Primary Examiner(s): SYED, FARHAN M
• Application Number:
  US11/001,404
• Time in Patent Office: 3,619 Days
• US Class Current: 707/730
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

In a method of generating a search boost vector, a first set of content location identifiers for first content are provided. A second set of content location identifiers for second content are identified. The second content is related to the first content in accordance with one or more predefined criteria. A search boost vector is generated by determining weights corresponding to content location identifiers in the second set of content location identifiers. The search boost vector may also be generated by determining weights corresponding to content location identifiers in the first set of content location identifiers. The predefined criteria may include one or more additional content location identifiers that are co-visited with one or more content location identifiers for the first set of content location identifiers. Co-visits are by a community of users during an interval of time.

Claims

1. A computer-implemented method, comprising:

at a search engine system comprising one or more processors and memory storing one or more programs executed by the one or more processors;

providing a first set of content location identifiers for first content;

using the first set of content location identifiers to identify a second set of content location identifiers for second content, wherein the second content is related to the first content based on co-visits by individual users, in a community of users, within a time interval, to a respective first content location identifier in the first set of content location identifiers and a respective second content location identifier in the second set of content location identifiers;

generating a search boost vector having weights corresponding to content location identifiers in the second set of content location identifiers, wherein the weights are based at least in part on a frequency of user visits to corresponding content location identifiers in the second set of content location identifiers; and

providing the search boost vector to a query server, wherein the search boost vector is for use by the query server to boost content location identifiers in the second set when serving search results.

2. The method of claim 1, wherein one or more weights are based on a frequency of co-visitations by individual users in the community of users.

3. The method of claim 1, wherein, on average, content location identifiers in the second set visited by numbers of users in the community of users in a first range have lower weights in the search boost vector than content location identifiers in the second set visited by numbers of users in the community of users in a second range, wherein the first range corresponds to substantially larger numbers of users than the second range.

4. The method of claim 1, wherein the first content is associated with a first user;

and comprising;

repeating the first providing, using, generating, and second providing for each user in a plurality of users, wherein, for each repeat, the first content is associated with a respective user in the plurality of users, and a respective search boost vector is provided to the query server for each user in the plurality of users.

5. The method of claim 1, wherein one or more of the weights in the search boost vector are inversely proportional to a number of content location identifiers visited by the users in the community of users between a visit by an individual user to the respective first content location identifier and a co-visit by the individual user to the respective second content location identifier.

6. The method of claim 1, wherein generating the search boost vector includes determining weights corresponding to content location identifiers in the first set of content location identifiers.

7. The method of claim 1, wherein generating the search boost vector includes applying a temporal window function to one or more of the weights corresponding to one or more of the content location identifiers in a set selected from the group consisting of the first set of content location identifiers, the second set of content location identifiers, and the first and second sets of content location identifiers, the temporal window function having weighting factors that are a function of the time interval.

8. The method of claim 1, wherein the search boost vector is ranked and truncated so as to include only entries with rank values exceeding a rank threshold.

9. The method of claim 1, wherein generating the search boost vector includes determining a consistency factor for one or more of the content location identifiers in the second set of content location identifiers.

10. The method of claim 9, wherein the consistency factor includes an absolute number of co-visits by individual users to the one or more of the content location identifiers in the second set of content location identifiers and the one or more of the content location identifiers in the first set of content location identifiers.

11. The method of claim 1, wherein generating the search boost vector includes determining a respective uniqueness factor for each of a plurality of respective content location identifiers in the second set of content location identifiers, wherein determining the respective uniqueness factor for a respective content location identifier in the second set of content location identifiers comprises dividing a co-visitation frequency of a respective first content location identifier with the respective second content identifier by a visit frequency of the respective second content identifier.

12. The method of claim 11, wherein using the first set of content location identifiers to identify the second set of content location identifiers includes determining if the uniqueness factor of a respective content location identifier exceeds a predefined uniqueness factor threshold.

13. A computer readable storage medium storing one or more programs executed by a computer system, the one or more programs comprising:

instructions for providing a first set of content location identifiers for first content;

instructions for using the first set of content location identifiers to identify a second set of content location identifiers for second content, wherein the second content is related to the first content based on co-visits by individual users, in a community of users, within a time interval, to a respective first content location identifier in the first set of content location identifiers and a respective second content location identifier in the second set of content location identifiers;

instructions for generating a search boost vector having weights corresponding to content location identifiers in the second set of content location identifiers, wherein the weights are based at least in part on a frequency of user visits to corresponding content location identifiers in the second set of content location identifiers; and

instructions for providing the search boost vector to a query server, wherein the search boost vector is for use by the query server to boost content location identifiers in the second set when serving search results.

14. The computer readable storage medium of claim 13, wherein one or more weights are based on a frequency of co-visitations by individual users in the community of users.

15. The computer readable storage medium of claim 13, wherein, on average, content location identifiers in the second set visited by numbers of users in the community of users in a first range have lower weights in the search boost vector than content location identifiers in the second set visited by numbers of users in the community of users in a second range, wherein the first range corresponds to substantially larger numbers of users than the second range.

16. The computer readable storage medium of claim 13, wherein the first content is associated with a first user;

and comprising;

instructions for repeating the first providing, using, generating, and second providing for each user in a plurality of users, wherein, for each repeat, the first content is associated with a respective user in the plurality of users, and a respective search boost vector is provided to the query server for each user in the plurality of users.

17. The computer readable storage medium of claim 13, wherein one or more weights are inversely proportional to a number of content location identifiers visited by the users in the community of users between a visit by an individual user to the respective first content location identifier and a co-visit by the individual user to the respective second content location identifier.

18. The computer readable storage medium of claim 13, wherein the instructions for generating the search boost vector include instructions for determining weights corresponding to content location identifiers in the first set of content location identifiers.

19. The computer readable storage medium of claim 13, wherein the instructions for generating the search boost vector include instructions for applying a temporal window function to one or more of the weights corresponding to one or more of the content location identifiers in a set selected from the group consisting of the first set of content location identifiers, the second set of content location identifiers and the first and second sets of content location identifiers, the temporal window function having weighting factors that are a function of the time interval.

20. The computer readable storage medium of claim 13, wherein the search boost vector is ranked and truncated so as to include only entries with rank values exceeding a rank threshold.

21. The computer readable storage medium of claim 13, wherein the instructions for generating the search boost vector include instructions for determining a consistency factor for one or more of the content location identifiers in the second set of content location identifiers.

22. The computer readable storage medium of claim 21, wherein the consistency factor includes an absolute number of co-visits by individual users to the one or more of the content location identifiers in the second set of content location identifiers and the one or more of the content location identifiers in the first set of content location identifiers.

23. The computer readable storage medium of claim 13, wherein the instructions for generating the search boost vector include instructions for determining a respective uniqueness factor for each of a plurality of respective content location identifiers in the second set of content location identifiers, wherein determining the respective uniqueness factor for a respective content location identifier in the second set of content location identifiers comprises dividing a co-visitation frequency of a respective first content location identifier with the respective second content identifier by a visit frequency of the respective second content identifier.

24. The computer readable storage medium of claim 23, wherein the instructions for using the first set of content location identifiers to identify the second set of content location identifiers include determining if the uniqueness factor for a respective content location identifier exceeds a predefined uniqueness factor threshold.

25. A computer, comprising:

a main memory;

a processor; and

a program, stored in the main memory and executed by the processor, the program including;

instructions for providing a first set of content location identifiers for first content;

instructions for using the first set of content location identifiers to identify a second set of content location identifiers for second content, wherein the second content is related to the first content based on co-visits by individual users, in a community of users, within a time interval, to a respective first content location identifier in the first set of content location identifiers and a respective second content location identifier in the second set of content location identifiers;

instructions for generating a search boost vector having weights corresponding to content location identifiers in the second set of content location identifiers, wherein the weights are based at least in part on frequency of user visits to corresponding content locations; and

instructions for providing the search boost vector to a query server, wherein the search boost vector is for use by the query server to boost content location identifiers in the second set when serving search results.

26. The computer of claim 25, wherein one or more weights are based on a frequency of co-visitations by individual users in the community of users.

27. The computer of claim 25, wherein, on average, content location identifiers in the second set visited by numbers of users in the community of users in a first range have lower weights in the search boost vector than content location identifiers in the second set visited by numbers of users in the community of users in a second range, wherein the first range corresponds to substantially larger numbers of users than the second range.

28. The computer of claim 25, wherein the first content is associated with a first user;

and comprising;

instructions for repeating the first providing, using, generating, and second providing for each user in a plurality of users, wherein, for each repeat, the first content is associated with a respective user in the plurality of users, and a respective search boost vector is provided to the query server for each user in the plurality of users.

29. The computer of claim 25, wherein one or more weights are inversely proportional to a number of content location identifiers visited by the users in the community of users between a visit by an individual user to the respective first content location identifier and a co-visit by the individual user to the respective second content location identifier.

30. The computer of claim 25, wherein the instructions for generating the search boost vector include instructions for determining weights corresponding to content location identifiers in the first set of content location identifiers.

31. The computer of claim 25, wherein the instructions for generating the search boost vector include instructions for applying a temporal window function to one or more of the weights corresponding to one or more of the content location identifiers in a set selected from the group consisting of the first set of content location identifiers, the second set of content location identifiers, and the first and second sets of content location identifiers, the temporal window function having weighting factors that are a function of the time interval.

32. The computer of claim 25, wherein the search boost vector is ranked and truncated so as to include only entries with rank values exceeding a rank threshold.

33. The computer of claim 25, wherein the instructions for generating the search boost vector include instructions for determining a consistency factor for one or more of the content location identifiers in the second set of content location identifiers.

34. The computer of claim 33, wherein the consistency factor includes an absolute number of co-visits by individual users to the one or more of the content location identifiers in the second set of content location identifiers and the one or more of the content location identifiers in the first set of content location identifiers.

35. The computer of claim 25, wherein the instructions for generating the search boost vector include instructions for determining a respective uniqueness factor for each of a plurality of respective content location identifiers in the second set of content location identifiers, wherein determining the respective uniqueness factor for a respective content location identifier in the second set of content location identifiers comprises dividing a co-visitation frequency of a respective first content location identifier with the respective second content identifier by a visit frequency of the respective second content identifier.

36. The computer of claim 35, wherein the instructions for using the first set of content location identifiers to identify the second set of content location identifiers include instructions for determining if the uniqueness factor for a respective content location identifier exceeds a predefined uniqueness factor threshold.

37. A computer-implemented method, comprising:

at a search engine system comprising one or more processors and memory storing one or more programs executed by the one or more processors;

receiving a search query from a user, wherein the user is a member of a community of users;

identifying search results including at least a portion of a second set of content location identifiers for second content, wherein the second set of content location identifiers is identified using a first set of content location identifiers for first content;

boosting at least a subset of the search results using a search boost vector to produce boosted search results,

the search boost vector having a plurality of weights, wherein a respective weight of the search boost vector is used to boost a respective content location identifier that is in the subset of the search results and that is also in the second set of content location identifiers,

wherein the respective weight is based, at least in part, on (A) co-visits by the user within a time interval to (i) the respective content location identifier in the second set of content location identifiers and (ii) a respective content location identifier in the first set of content location identifiers, and (B) a frequency of user visits to the respective content location identifier in the second set of content location identifiers; and

providing the boosted search results to the user.

38. The method of claim 37, further comprising:

selecting the search boost vector from a plurality of search boost vectors based on at least one of;

user preference information, user profile information, and content site profiles.

39. A computer readable storage medium storing one or more programs executed by a computer system, the one or more programs comprising:

instructions for receiving a search query from a user, wherein the user is a member of a community of users;

instructions for identifying search results including at least a portion of a second set of content location identifiers for second content, wherein the second set of content location identifiers is identified using a first set of content location identifiers for first content;

instructions for boosting at least a subset of the search results using a search boost vector to produce boosted search results,

the search boost vector having a plurality of weights, wherein a respective weight of the search boost vector is used to boost a respective content location identifier that is in the subset of the search results and that is also in the second set of content location identifiers,

wherein the respective weight is based, at least in part, on (A) co-visits by the user within a time interval to (i) the respective content location identifier in the second set of content location identifiers and (ii) a respective content location identifier in the first set of content location identifiers, and (B) a frequency of user visits to the respective content location identifier in the second set of content location identifiers; and

instructions for providing the boosted search results to the user.

40. The computer readable storage medium of claim 39, further comprising:

instructions for selecting the search boost vector from a plurality of search boost vectors based on at least one of;

user preference information, user profile information, and content site profiles.

41. A computer, comprising:

a main memory;

a processor; and

a program, stored in the main memory and executed by the processor, the program including;

instructions for receiving a search query from a user, wherein the user is a member of a community of users;

instructions for identifying search results including at least a portion of a second set of content location identifiers for second content, wherein the second set of content location identifiers is identified using a first set of content location identifiers for first content;

instructions for boosting at least a subset of the search results using a search boost vector to produce boosted search results,

the search boost vector having a plurality of weights, wherein a respective weight of the search boost vector is used to boost a respective content location identifier that is in the subset of the search results and that is also in the second set of content location identifiers,

wherein the respective weight is based, at least in part, on (A) co-visits by the user within a time interval to (i) the respective content location identifier in the second set of content location identifiers and (ii) a respective content location identifier in the first set of content location identifiers, and (B) a frequency of user visits to the respective content location identifier in the second set of content location identifiers; and

instructions for providing the boosted search results to the user.

42. The computer of claim 41, further comprising:

instructions for selecting the search boost vector from a plurality of search boost vectors based on at least one of;

user preference information, user profile information, and content site profiles.