Recommendation of related electronic assets based on user search behavior

• US 7,752,201 B2
• Filed: 05/10/2007
• Issued: 07/06/2010
• Est. Priority Date: 05/10/2007
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Suzuki, Keiichiro, Anderson, Blake Edward
• Primary Examiner(s): Stevens; Robert
• Application Number:
  US11/803,018
• Publication Number:
  US 20080281808A1
• Time in Patent Office: 1,153 Days
• Field of Search: 707/2, 707/5, 707/713, 707/731, 707/758
• US Class Current: 707/731
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Recommendation of related assets based on user search behavior is provided. Relevance classifications assigned to search results in association with search strings that produced the search results may be used for providing a list of one or more searchable assets that may be relevant to a given search query. All pairings of two acceptable assets contained in a given search session may be marked as “See Also” candidates. If a given pairing appears in a search session more than a configurable threshold, the assets in the pairing may be established as “See Also” relationships. Subsequently, when a user conducts a search that produces a search result that is a member of a pairing established as a “See Also” relationship, the other asset in the pairing may be recommended to the searching user within the destination search result page after the user selects the search result for viewing.

Claims

1. A method of providing related search assets to an obtained search asset, the method comprising:

providing a relevance classification for each search asset in a plurality of sessions,

wherein for each search asset in the plurality of sessions, the relevance classification for the search asset is based on a relevance of the search asset to a search query that was used to locate the search asset, and

wherein a search asset is an acceptable search asset when the search asset has a relevance classification indicating that the search asset is a correct response to the search query;

identifying, by a computer, a set of pairings, wherein for each session in the plurality of sessions, the set of pairings includes each pairing of acceptable search assets in the session; and

when a given pairing appears in the set of pairings more than a configurable threshold count, establishing the two search assets comprising the given pairing as recommended related search assets.

2. The method of claim 1, further comprising when a search for one or more search assets obtains one of the two search assets comprising the given pairing and when the given pairing has been established as recommended related search assets, recommending the other of the two search assets comprising the given pairing to a searching user.

3. The method of claim 2, further comprising recommending the other of the two search assets comprising the given pairing to the searching user in a “

See Also”

list of recommended search assets.

4. The method of claim 1, further comprising maintaining a “

See Also”

data record for each pairing in the set of pairings.

5. The method of claim 4, further comprising maintaining a count in the “

See Also”

data record for the given pairing, the count indicating a number of users associated with the given pairing.

6. The method of claim 5, further comprising:

maintaining a “

See Also User”

data record for each of the users associated with the given pairing;

relating the “

See Also User”

data records for the users associated with the given pairing to the “

See Also”

data record for the given pairing; and

wherein an appearance of the given pairing is not counted toward the configurable threshold count when the appearance is associated with a particular user and when another appearance of the given pairing is associated with the particular user and is counted toward the configurable threshold count, thereby preventing two search assets from being established as recommended related search assets based on a user behavior of a single user.

7. The method of claim 1, further comprising:

prior to providing the relevance classification for each search asset in the plurality of sessions;

calculating one or more relevance factors for each search asset in the plurality of sessions; and

providing the relevance classification for each search asset based on the one or more relevance factors calculated for each search asset in the plurality of sessions.

8. The method of claim 7, wherein providing the relevance classification for each search asset based on the one or more relevance factors calculated for each search asset included in the plurality includes analyzing user behavior associated with a plurality of search assets for which explicit feedback is provided.

9. The method of claim 8, wherein analyzing user behavior associated with the plurality of search assets for which explicit feedback is provided includes determining a user behavior based on one or more relevance factors associated with a search asset for which explicit feedback is provided and for which the search asset is acceptable to a searching user.

10. The method of claim 9, wherein providing the relevance classification for each search asset based on the one or more relevance factors calculated for each search asset included in the plurality of sessions includes applying the user behavior to one or more relevance factors calculated for each search asset included in the plurality of sessions for classifying one or more search assets included in the plurality of sessions for which no explicit feedback is provided as acceptable.

11. A computer storage medium containing computer readable instructions for providing related search results to an obtained search result, the computer readable instructions, when executed, causing a computer to:

group search results in a plurality of search results into a plurality of sessions, each search result in the plurality of search results being an asset identified in response to a search query in a plurality of search queries;

for each session in the plurality of sessions;

for each search result in the session;

calculate one or more relevance factors for the search result, wherein the relevance factors for the search result indicate user satisfaction with the search result; and

provide a relevance classification for the search result based on the relevance factors for the search result, wherein the search result is an acceptable search result when the search result has a relevance classification that indicates that the search result is a correct response to a search query used to locate the search result;

identify a set of pairings, wherein for each session in the plurality of sessions, the set of pairings includes each pairing of acceptable search results included within the session; and

when a given pairing appears in the set of pairings more than a configurable threshold count, establish the two search results comprising the given pairing as recommended related search results.

12. The computer storage medium of claim 11, wherein when a search for one or more search results obtains one of the two search results comprising the given pairing and when the given pairing has been established as recommended related search results, the computer readable instructions, when executed, cause the computer to recommend the other of the two search results comprising the given pairing to a searching user.

13. The computer storage medium of claim 12, further comprising recommending the other of the two search results comprising the given pairing to the searching user in a “

See Also”

list of recommended search results.

14. The computer storage medium of claim 11, further comprising maintaining a “

See Also”

data record for each pairing in the set of pairings.

15. The computer storage medium of claim 14, further comprising maintaining a count in the “

See Also”

data record for the given pairing, the count indicating a number of users associated with the given pairing.

16. The computer storage medium of claim 15, wherein the computer readable instructions, when executed, further cause the computer to:

maintain a “

See Also User”

data record for each of the users associated with the given pairing;

relate the “

See Also User”

data record for the users associated with the given pairing to the “

See Also”

data record for the given pairing;

wherein an appearance of the given pairing is not counted toward the configurable threshold count when the appearance is associated with a particular user and when another appearance of the given pairing is associated with the particular user and is counted toward to configurable threshold count, thereby preventing two search results from being established as recommended related search results based on a user behavior of a single user.

17. The computer storage medium of claim 11, wherein providing a relevance classification for each search result included in the session includes analyzing user behavior associated with the plurality of search results for which explicit feedback is provided.

18. The computer readable storage medium of claim 17, wherein analyzing user behavior associated with the plurality of search results for which explicit feedback is provided includes determining a user behavior based on one or more relevance factors associated with a search result for which explicit feedback is provided and for which the search result is acceptable to a searching user.

19. The computer storage medium of claim 18, wherein providing the relevance classification for each search result in the session based on a relevance of each search result to a search query that was used to locate each search result includes applying the user behavior to one or more relevance factors calculated for each search result in the session for classifying one or more search results in the session for which no explicit feedback is provided as acceptable.

20. A computer storage medium containing computer readable instructions for providing related search results to an obtained search result, the computer readable instructions, when executed by a computer, cause the computer to:

group search results in a plurality of search results into a plurality of sessions, each search result in the plurality of search results being an asset identified in response to a search query in a plurality of search queries;

for each session in a plurality of sessions;

for each search result in the session;

calculate one or more relevance factors for the search result, wherein the relevance factors for the search result indicate user satisfaction with the search result;

provide a relevance classification for the search result based on the relevance factors for the search result and based on user behavior associated with search results in the session for which explicit feedback is provided, wherein the search result is an acceptable search result when the search result has a relevance classification that indicates that the search result is a correct response to a search query used to locate the search result;

find a plurality of sessions having more than one search result classified as an acceptable search result;

identify a set of pairings, wherein for each session in the plurality sessions having more than one search result classified as an acceptable search result, the set of pairings includes each pairing of acceptable search results included within the session;

maintain a user data record for each user associated with a given pairing in the set of pairings;

maintain a “

See Also”

data record for the given pairing, the “

See Also”

data record for the given pairing comprising a count indicating a number of the users associated with the given pairing;

when the count in the “

See Also”

data record for the given pairing is more than a configurable threshold count, establish the two search results comprising the given pairing as recommended related search results; and

when a search for one or more search results obtains one of the two search results comprising the given pairing and when the given pairing has been established as recommended related search results, recommending the other of the two search results comprising the given pairing to a searching user.