QUERY SUGGESTION FOR E-COMMERCE SITES

• US 20120036123A1
• Filed: 07/28/2011
• Published: 02/09/2012
• Est. Priority Date: 07/30/2010
• Status: Active Grant

• Current Assignee: eBay Inc.
• Original Assignee: eBay Inc.
• Inventors: Parikh, Nishith, Singh, Gyanit, Sundaresan, Neelakantan, Johnson, Brian Scott, Hasan, Mohammad Al, Khurana, Udayan
• Granted Patent: US 8,954,422 B2
• Time in Patent Office: Days
• US Class Current: 707/723
• CPC Class Codes: G06F 16/24 Querying
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/3325 Reformulation based on results of preceding query
• CPC Class Codes: G06F 16/95 Retrieval from the web
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0625 Directed, with specific intent or strategy

Abstract

Methods, articles of manufacture and a system for providing query suggestions using a query log that includes a number of user sessions. The sessions comprise training data including a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. The query log is cleaned and normalized. From the cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets is generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the level are included as elements of the set of query suggestions that meet the predetermined assurance level. The set of query suggestions are mixed and ranked in accordance with a user behavior sought to be optimized.

Claims

1. A method for providing query suggestions for a user comprising:

using a computer processor and storage,

receiving a query log comprising a number of user sessions, the sessions including a plurality of sets of queries, some of the sets of queries including query transitions, followed by a purchase related event;

cleaning and normalizing the query log;

generating from the cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets, the transition scores comprising ordered query pairs;

building a set of query suggestions from the ordered query pairs;

computing similarity scores of at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level;

including as elements of the set of query suggestions that meet the predetermined assurance level; and

mixing and ranking the set of query suggestions in accordance with a user behavior that is to be optimized.

2. The method of claim 1, the stationary scores and the transition scores are computed using one of a popularity score and a purchase efficiency score.

3. The method of claim 2 wherein the purchase efficiency score is the linear combination of a count of different purchase related events.

4. The method of claim 1 wherein the similarity is a semantic similarity.

5. The method of claim 1 further including normalizing the scores for the set of query suggestions.

6. The method of claim 5 wherein the normalizing comprises dividing the number of query transitions by the number of users that practice the query transition.

7. The method of claim 1 wherein the suggestion mixing comprises one of refinement mixing and related mixing.

8. The method of claim 7 wherein the refinement mixing includes adding one or more terms in one of the sets of queries to constrain a search that uses the query suggestion.

9. The method of claim 1 wherein the ranking comprises combining a popularity score and a purchase efficiency score utilizing different weights.

10. The method of claim 1 wherein the set of query suggestions includes a first query suggestion and further including cleaning the set of query suggestions by deleting the first query suggestion based on a predetermined probability of returning a null search results page on an e-commerce site.

11. A non-transitory computer-readable storage medium having embedded therein a set of instructions which, when executed by one or more processors of a computer, causes the computer to execute the following operations:

receiving a query log comprising a number of user sessions, the sessions comprising a plurality of sets of queries, some of the sets of queries including query transitions, followed by a purchase related event;

cleaning and normalizing the query log;

generating from the cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets, the transition scores comprising ordered query pairs;

building a set of query suggestions from the ordered query pairs;

computing similarity scores of at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level;

including as elements of the set of query suggestions that meet the predetermined assurance level; and

mixing and ranking the set of query suggestions in accordance with a user behavior that is to be optimized.

12. The computer readable storage medium of claim 11 wherein stationary scores and the transition scores computed using one of a popularity score and a purchase efficiency score.

13. The computer readable storage medium of claim 12 wherein the purchase efficiency score is a linear combination of a count of different purchase related events.

14. The computer readable storage medium of claim 11 wherein the similarity is a semantic similarity.

15. The computer readable storage medium of claim 11, the operations further including normalizing the scores for the set of query suggestions.

16. The computer readable storage medium of claim 15 wherein the normalizing comprises dividing the number of query transitions by the number of users that practice the query transition.

17. The computer readable storage medium of claim 11 wherein the suggestion mixing comprises one of refinement mixing and related mixing.

18. The computer readable storage medium of claim 17 wherein refinement mixing includes adding one or more terms in one of the sets of queries to constrain a search that uses the query suggestion.

19. The computer readable storage medium of claim 11 wherein ranking comprises combining a popularity score and a purchase efficiency score by applying different weights to the popularity score and to the purchase efficiency score.

20. The computer readable storage medium of claim 11 wherein the set of query suggestions includes a first query suggestion and further including cleaning the query suggestions by deleting the first query suggestion based on a predetermined probability of returning a null search results page on an e-commerce site.

21. A system for providing query suggestions for a user comprising:

a computer processor and storage configured to;

receive a query log comprising a number of user sessions, the sessions comprising training data including a sequence of a plurality of sets of queries, some of the sets of queries including query transitions, followed by a purchase related event;

clean and normalize the query log using a cleaning and normalizing module;

generate from the cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets, the transition scores comprising ordered query pairs;

build a set of query suggestions from the ordered query pairs;

compute similarity scores of at least some of the set of query suggestions to determine whether individual ones of the at least some of the sets of query suggestions meet a predetermined assurance level;

include as elements of the set of query suggestions that meet the predetermined assurance level; and

mix and rank the set of query suggestions in accordance with a user behavior that is to be optimized.

22. A method of generating suggestions for search, the method comprising:

using a computer processor and storage, mining user query activity from user activity history logs;

building a graph from a user query activity history log which includes at least one of searches, bins, bids, ask seller a question, watches, views, and offers, wherein

at least some queries are seen in the user activity history log as a node on the graph,

edges connecting the at least some queries are based on follow counts of queries in user sessions, and

popularity of queries and user behavior in terms of engagement are properties seen as edges of the graph; and

pruning the graph, based on edge properties to obtain the best neighbors for every query which are likely to lead the user to purchases.

23. The method of claim 22, the method employing Hadoop clustering.

24. The method of claim 22, the method employing noise removal techniques including one of historical inventory count, current inventory count, and Porter'"'"'s stemmer.

25. A non-transitory computer-readable storage medium having embedded therein a set of instructions which, when executed by one or more processors of a computer, causes the computer to execute the following operations:

mining user query activity from user activity history logs;

building a graph from a user query activity history log which includes at least one of searches, bins, bids, ask seller a question, watches, views, and offers, wherein

at least some queries are seen in the user activity history log as a node on the graph,

edges connecting the at least some queries are based on follow counts of queries in user sessions, and

popularity of queries and user behavior in terms of engagement are properties seen as edges of the graph; and

pruning the graph, based on edge properties to obtain the best neighbors for every query which are likely to lead the user to purchases.