Recent interest based relevance scoring

• US 8,972,391 B1
• Filed: 10/02/2009
• Issued: 03/03/2015
• Est. Priority Date: 10/02/2009
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: McDonnell, Philip, Jeh, Glen M., Haveliwala, Taher H., Kurzion, Yair
• Primary Examiner(s): KINSAUL, DANIEL W
• Application Number:
  US12/572,739
• Time in Patent Office: 1,978 Days
• US Class Current: 707/727
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/337 Profile generation, learning or modification
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

A computer-implemented method for processing query information includes receiving prior queries followed by a current query, the prior and current queries being received within an activity period an originating with a search requester. The method also includes receiving a plurality of search results based on the current query. Each search result identifying a search result document, each respective search result document being associated with a query specific score indicating a relevance of the document to the current query. The method also includes determining a first category based, at least in part, on the prior queries. The method also includes identifying a plurality of prior activity periods of other search requesters, each prior activity period containing a prior activity query where the prior activity query matches the current query, and where the prior activity period indicates the same first category. The method also includes obtaining category based selection statistics of the other requesters search results based on the last queries of the other activity periods. The method also includes obtaining general selection statistics of a more general population of requesters based on the current query. The method also includes generating adjusted scores for the search result documents by adjusting the respective scores based on the query specific score and the category specific score. The method also includes ranking the search result documents according to the respective adjusted scores.

Claims

1. A non-transitory computer readable storage medium having a computer program stored thereon, the program comprising instructions that, when executed by data processing apparatus, cause the data processing apparatus to perform operations comprising:

analyzing a recent search activity period of a user to determine a short-term category of interest of the user, wherein the analyzing comprises comparing queries submitted by the user during the recent search activity period with categories of selected search results that were selected by the user during the recent search activity period;

obtaining a plurality of search results responsive to a query, each of the plurality of search results having a respective score;

for each particular search result of a first plurality of the search results;

calculating a category selection value for the particular search result, wherein the category selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for the short-term category of interest;

calculating a general selection value for the particular search result, wherein the general selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for any category of interest;

calculating a category relevance for the particular search result, wherein the respective category relevance for the particular search result is based on a difference between the respective category selection value for the search result and the respective general selection value for the particular search result;

selecting one or more search results of the plurality of search results, each selected search result being selected based on the selected search result having a category relevance that exceeds a threshold;

adjusting the respective score for each of the selected search results based on, at least, the category selection value for each selected search result and the general selection value for each selected search result; and

ranking the search results of the plurality of search results according to the respective adjusted scores for the selected search results and the respective scores for search results that were not selected.

2. The non-transitory computer-readable storage medium of claim 1, wherein analyzing the recent search activity period of the user comprises determining an amount of time elapsed between selection of a first search result in the recent search activity period and selection of a second search result in the recent search activity period.

3. The non-transitory computer-readable storage medium of claim 1, wherein each of one or more of the selections of the particular search result is weighted by a respective weight based on an amount of time that a user viewed a document referred to by the particular search result.

4. The non-transitory computer-readable storage medium of claim 1, wherein the category relevance for each selected search result is statistically significant.

5. The non-transitory computer-readable storage medium of claim 1, wherein adjusting the respective score for a particular selected search result comprises combining the category relevance for the particular selected search result with an information retrieval score of the particular selected search result.

6. The non-transitory computer-readable storage medium of claim 5, wherein combining the category relevance for the selected search result with the information retrieval score of the particular selected search result comprises:

multiplying the category relevance for the selected search result by the information retrieval score of the particular selected search result, or adding the category relevance for the particular selected search result by the information retrieval score of the particular selected search result.

7. The non-transitory computer-readable storage medium of claim 1, wherein determining the short-term category of interest comprises determining the short-term category of interest based on, at least, user specific information.

8. The non-transitory computer-readable storage medium of claim 7, wherein the user specific information includes at least one of demographic information of the user and location information of the user.

9. The non-transitory computer readable storage medium of claim 1, wherein:

the category selection value for the particular search result is a first click-through rate for the particular search result when the particular search result is presented in response to the query and to a first group of users having an interest that matches the short-term category of interest;

the general selection value for the particular search result is a second click-through rare for the particular search result when the particular search result is presented in response to the query and to a second group of users that have any category of interest, the second group of users having at least one user that is not included in the first group of users; and

the category relevance is based on a difference between the first click-through rate and the second click-through rate.

10. A computer-implemented method comprising:

analyzing a recent search activity period of a user to determine a short-term category of interest of the user, wherein the analyzing comprises comparing queries submitted by the user during the recent search activity period with categories of selected search results that were selected by the user during the recent search activity period;

obtaining a plurality of search results responsive to a query, each of the plurality of search results having a respective score;

for each particular search result of a first plurality of the search results;

calculating a category selection value for the particular search result, wherein the category selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for the short-term category of interest;

calculating a general selection value for the particular search result, wherein the general selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for any category of interest;

calculating a category relevance for the particular search result, wherein the respective category relevance for the particular search result is based on a difference between the respective category selection value for the search result and the respective general selection value for the particular search result;

selecting a one or more search results of the first plurality of search results, each selected search result being selected based on the selected search result having a category relevance that exceeds a threshold;

adjusting the respective score for each of the selected search results based on, at least, the category selection value for each selected search result and the general selection value for each selected search result; and

ranking the search results of the plurality of search results according to the respective adjusted scores for the selected search results and the respective scores for search results that were not selected.

11. The method of claim 10, wherein analyzing the recent search activity period of the user comprises determining an amount of time elapsed between selection of a first search result in the recent search activity period and selection of a second search result in the recent search activity period.

12. The method of claim 10, wherein each of one or more of the selections of the particular search result is weighted by a respective weight based on an amount of time that a user viewed a document referred to by the particular search result.

13. The method of claim 10, wherein the category relevance for each selected search result is statistically significant.

14. The method of claim 10, wherein adjusting the respective score for a particular selected search result comprises combining the category relevance for the particular selected search result with an information retrieval score of the particular selected search result.

15. The method of claim 14, wherein combining the category relevance for the selected search result with the information retrieval score of the particular selected search result comprises:

multiplying the category relevance for the particular selected search result by the information retrieval score of the particular selected search result, or adding the category relevance the particular selected search result by the information retrieval score of the particular selected search result.

16. The method of claim 10, wherein determining the short-term category of interest comprises determining the short-term category if interest based on, at least, user specific information.

17. The method of claim 16, wherein the user specific information includes at least one of demographic information of the user and location information of the user.

18. The method of claim 10, wherein:

the category selection value for the particular search result is a first click-through rate for the particular search result when the particular search result is presented in response to the query and to a first group of users having an interest that matches the short-term category of interest;

the general selection value for the particular search result is a second click-through rare for the particular search result when the particular search result is presented in response to the query and to a second group of users that have any category of interest, the second group of users having at least one user that is not included in the first group of users; and

the category relevance is based on a difference between the first click-through rate and the second click-through rate.

19. A system comprising:

data processing apparatus programmed to perform operations comprising;

analyzing a recent search activity period of a user to determine a short-term category of interest of the user, wherein the analyzing comprises comparing queries submitted by the user during the recent search activity period with categories of selected search results that were selected by the user during the recent search activity period;

obtaining a plurality of search results responsive to a query, each of the plurality of search results having a respective score;

for each particular search result of a first plurality of the search results;

calculating a category selection value for the particular search result, wherein the category selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for the short-term category of interest;

calculating a general selection value for the particular search result, wherein the general selection value for the particular search result is based on a measure of a count of selections of the particular search result as a portion of combined selection counts for search results responsive to the query for a plurality of users for any category of interest;

calculating a category relevance for the particular search result, wherein the respective category relevance for the particular search result is based on a difference between the respective category selection value for the search result and the respective general selection value for the particular search result;

selecting a one or more search results of the first plurality of search results, each selected search result being selected based on the selected search result having a category relevance that exceeds a threshold;

adjusting the respective score for each of the selected search results based on, at least, the category selection value for each selected search result and the general selection value for each selected search result; and

ranking the search results of the plurality of search results according to the respective adjusted scores for the selected search results and the respective scores for search results that were not selected.

20. The system of claim 19, wherein analyzing the recent search activity period of the user comprises determining an amount of time elapsed between selection of a first search result in the recent search activity period and selection of a second search result in the recent search activity period.

21. The system of claim 19, wherein each of one or more of the selections of the particular search result is weighted by a respective weight based on an amount of time that a user viewed a document referred to by the particular search result.

22. The system of claim 19, wherein the category relevance for each selected search result is statistically significant.

23. The system of claim 19, wherein adjusting the respective score for a particular selected search result comprises combining the category relevance for the selected search result with an information retrieval score of the particular selected search result.

24. The system of claim 23, wherein combining the category relevance for the particular selected search result with the information retrieval score of the particular selected search result comprises:

multiplying the category relevance for the particular selected search result by the information retrieval score of the particular selected first search result, or adding the category relevance for the selected search result by the information retrieval score of the particular selected search result.

25. The system of claim 19, wherein determining the short-term category of interest comprises determining the short-term category of interest based on, at least, user specific information.

26. The system of claim 25, wherein the user specific information includes at least one of demographic information of the user and location information of the user.

27. The system of claim 19, wherein:

the category selection value for the particular search result is a first click-through rate for the particular search result when the particular search result is presented in response to the query and to a first group of users having an interest that matches the short-term category of interest;

the general selection value for the particular search result is a second click-through rare for the particular search result when the particular search result is presented in response to the query and to a second group of users that have any category of interest, the second group of users having at least one user that is not included in the first group of users; and

the category relevance is based on a difference between the first click-through rate and the second click-through rate.