Community search optimization

• US 8,005,823 B1
• Filed: 03/28/2007
• Issued: 08/23/2011
• Est. Priority Date: 03/28/2007
• Status: Expired due to Fees

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Scofield, Christopher L., Pope, Elmore Eugene, Nordstrom, Paul G., Marshall, Brad E.
• Primary Examiner(s): Ali; Mohammad
• Assistant Examiner(s): Tran; Bao G
• Application Number:
  US11/692,787
• Time in Patent Office: 1,609 Days
• Field of Search: 707/3, 707/5, 707/10, 707/2, 707/721, 707/713, 707/728, 707/731, 707/759, 707/769, 705/37, 705/26, 709/203
• US Class Current: 707/721
• CPC Class Codes: G06F 16/2455 Query execution

Abstract

A community search system may utilize the results of previous queries from community members and user feedback associated with those results to provide results for subsequent, similar queries from within the same community. Thus, by modifying and combining query results based, in part, on user feedback regarding previous results for similar queries, the trial and error of the many may improve the results for an individual. Users may be grouped into communities whose members are associated by common interests or characteristics. Feedback from one community member may be used to modify the query results for other community members. By analyzing user feedback with the community regarding query results, negative, incorrect or unnecessary results may be omitted or removed from subsequent results for similar queries.

Claims

1. A system, comprising:

a processor; and

a memory coupled to the processor, wherein the memory comprises program instructions executable by the processor to;

group users into a plurality of distinct communities according to one or more grouping criteria, such that for each particular one of the communities, members of the particular community share at least one grouping criterion in common, and wherein membership of at least some of the communities does not completely overlap;

store indications of membership of each of the communities;

receive a query from a member of one or more of the communities, and in response to the query;

obtain results of the query;

determine one or more of the plurality of communities to associate with the query, wherein the determining comprises comparing the query to queries entered by other members of the one or more of the plurality of communities such that the member from which the query is received is a member of the determined one or more communities, and the determined one or more communities are distinct from others of the plurality of communities;

identify other results from only the determined one or more communities, wherein the other results are for at least one other query from at least one other member of the determined one or more communities, and wherein the other results reflect user feedback from the at least one other member of the determined one or more communities;

compare the results to the other results from the determined one or more communities to determine a measure of similarity between the results and the other results; and

modify the results according to the other results in response to determining that the measure of similarity is above a predetermined threshold, wherein the results are not modified according to the other results if the measure of similarity is not above the predetermined threshold.

2. The system of claim 1, wherein the program instructions are further executable to receive user feedback regarding the modified results from the community member, wherein the feedback specifies a level of correctness for at least one of the results, and wherein the feedback comprises feedback explicitly indicated by the community member selecting a positive or negative indication from among a plurality of indications.

3. The system of claim 2, wherein the program instructions are further executable to further modify the results according to the received user feedback.

4. The system of claim 3, wherein to modify the results according to the received user feedback, the program instructions are further executable to modify a ranking of the results, wherein one of the results is ranked higher than another of the results if the received user feedback specifies a higher level of correctness for the one of the results than a level of correctness specified for the another of the first results.

5. The system of claim 2, wherein the program instructions are further executable to compare the received user feedback with user feedback from the another community member for at least one of the other results to determine a level of similarity between the feedback received from the community member and the feedback received from the another community member.

6. The system of claim 1, wherein to modify the results the program instructions are further executable to aggregate the results and the other results to produce aggregated results, wherein the aggregated results include at least one result from the other results that was not in the results obtained for the query.

7. The system of claim 6, wherein to aggregate the results with the other results, the program instructions are executable to determine one or more of the other results for which the user feedback specifies a level of correctness above a pre-defined threshold, and include those results in the aggregated results.

8. The system of claim 6, wherein to aggregate the results with the other results, the program instructions are executable to exclude one or more of the results from inclusion in the aggregated results in response to determining that the user feedback for those results specifies a level of correctness below a pre-defined threshold.

9. The system of claim 1, wherein to determine a measure of similarity between the results and the other results the program instructions are further executable to calculate a ratio of a number of the results in common with the other results to a total number of the results.

10. A computer-implemented method, comprising:

grouping users into a plurality of distinct communities according to one or more grouping criteria, such that for each particular one of the communities, members of the particular community share at least one grouping criterion in common, and wherein membership of at least some of the communities does not completely overlap;

storing indications of membership of each of the communities;

receiving a query from a member of one or more of the communities, and in response to the query;

obtaining results of the query;

determining one or more of the plurality of communities to associate with the query, wherein the determining comprises comparing the query to queries entered by other members of the one or more of the plurality of communities such that the member from which the query is received is a member of the determined one or more communities, and the determined one or more communities are distinct from others of the plurality of communities;

identifying other results from only the determined one or more communities, wherein the other results are for at least one other query from at least one other member of the determined one or more communities, and wherein the other results reflect user feedback from the at least one other member of the determined one or more communities;

comparing the results to the other results from the determined one or more communities to determine a measure of similarity between the results and the other results; and

modifying the results according to the other results in response to determining that the measure of similarity is above a predetermined threshold, wherein the results are not modified according to the other results if the measure of similarity is not above the predetermined threshold;

wherein each of said grouping, said storing, said receiving, said obtaining, said identifying, said comparing, and said modifying is performed by one or more computer systems, each comprising at least a processor and a memory.

11. The method of claim 10, further comprising receiving user feedback regarding the modified results from the community member, wherein the feedback specifies a level of correctness for at least one of the results, and wherein the feedback comprises feedback explicitly indicated by the community member selecting a positive or negative indication from among a plurality of indications.

12. The method of claim 11, further comprising modifying the results according to the received user feedback.

13. The method of claim 12, wherein said modifying comprises modifying a ranking of the results, wherein one of the results is ranked higher than another one of the results if the received user feedback specifies a higher level of correctness for the one of the results than a level of correctness specified for the another one of the results.

14. The method of claim 11, further comprising comparing the received user feedback with user feedback from the another community member specifying the correctness of at least one of the other results to determine a similarity measure corresponding to a level of similarity between the feedback received from the community member and the feedback received from the another community member.

15. The method of claim 10, wherein said modifying the results comprises aggregating the results and the other results to produce aggregated results, wherein the aggregated results include at least one result from the other results that was not in the results obtained for the query.

16. The method of claim 15, wherein said aggregating comprises:

determining one or more of the other results for which the user feedback specifies a level of correctness above a pre-defined threshold; and

including those results in the aggregated results.

17. The method of claim 15, wherein said aggregating comprises excluding one or more results from inclusion in the aggregated results in response to determining that the user feedback for those results specifies a level of correctness below a pre-defined threshold.

18. A computer-readable storage medium, comprising program instructions executable to implement:

grouping users into a plurality of distinct communities according to one or more grouping criteria, such that for each particular one of the communities, members of the particular community share at least one grouping criterion in common, and wherein membership of at least some of the communities does not completely overlap;

storing indications of membership of each of the communities;

receiving a query from a member of one or more of the communities, and in response to the query;

obtaining results of the query;

determining one or more of the plurality of communities to associate with the query, wherein the determining comprises comparing the query to queries entered by other members of the one or more of the plurality of communities such that the member from which the query is received is a member of the determined one or more communities, and the determined one or more communities are distinct from others of the plurality of communities;

identifying other results from only the determined one or more communities, wherein the other results are for at least one other query from at least one other member of the determined one or more communities, and wherein the other results reflect user feedback from the at least one other member of the determined one or more communities;

comparing the results to the other results from the determined one or more communities to determine a measure of similarity between the results and the other results; and

modifying the results according to the other results in response to determining that the measure of similarity is above a predetermined threshold, wherein the results are not modified according to the other results if the measure of similarity is not above the predetermined threshold.

19. The medium of claim 18, wherein the program instructions are further executable to implement receiving user feedback regarding the modified results from the community member, wherein the feedback specifies a level of correctness for at least one of the results, and wherein the feedback comprises feedback explicitly indicated by the community member selecting a positive or negative indication from among a plurality of indications.

20. The medium of claim 19, wherein the program instructions are further executable to implement modifying the results according to the received user feedback.

21. The medium of claim 20, wherein said modifying comprises modifying a ranking of the results, wherein one of the results is ranked higher than another one of the results if the received user feedback specifies a higher level of correctness for the one of the results than a level of correctness specified for the another one of the results.

22. The medium of claim 20, further comprising comparing the received user feedback with user feedback from the another community member specifying the correctness of at least one of the other results to determine a similarity measure corresponding to a level of similarity between the feedback received from the community member and the feedback received from the another community member.