Efficiently identifying the items most relevant to a current query based on items selected in connection with similar queries

• US 7,574,426 B1
• Filed: 03/31/2005
• Issued: 08/11/2009
• Est. Priority Date: 03/31/2005
• Status: Active Grant

• Current Assignee: A9.com Incorporated (Amazon.com, Inc.)
• Original Assignee: A9.com Incorporated (Amazon.com, Inc.)
• Inventors: Ortega, Ruben E.
• Primary Examiner(s): Breene; John E
• Assistant Examiner(s): Kerzhner; Aleksandr
• Application Number:
  US11/097,468
• Time in Patent Office: 1,594 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A facility for processing a query made up of one or more query terms is described. The facility receives the query in a search engine. In the search engine, the facility generates a search result that is made up of items. The generated search result includes items with which are associated one or more of the query terms of the received query. The generated search result further reflects the extent to which users have selected the items of the generated search result when the items of the generated search result have appeared in search results generated for queries having one or more query terms in common with the received query.

Claims

1. A method in a computing system for processing a query made up of one or more query terms to identify relevant items, comprising:

receiving the query to a search engine able to utilize a search index to generate a search result, the search index containing an entry for each item having both (1) one or more item matching fields that contain query terms having a direct association with the item and (2) one or more item selection fields that contain one or more query terms each repeated a number of times representing a level of relevance of each of the one or more query terms to the item, the level of relevance based at least in part upon previous selection actions performed by users with respect to the item, each of the users having submitted a previous query including at least one of the one or more query terms and having received a previous search result identifying the item;

determining an ordered set of items relating to the received query using the search index and the one or more query terms of the received query, an order of the ordered set of items being determined at least in part by generating a ranking score for each item in the ordered set of items, the ranking score being generated for each item in the ordered set of items by assigning a field score for each occurrence of one of the one or more query terms in one of the item selection and item information fields and summing the field scores for each item; and

generating a search result made up of the ordered set of items,

wherein the search engine is able to generate the search result without modifying operation of the search engine.

2. The method of claim 1 wherein the items making up the generated search result are ordered in a way that reflects the extent to which users select the items of the generated search result when the items of the generated search result appear in search results generated for queries having one or more query terms in common with the received query.

3. The method of claim 1 wherein the items making up the generated search result are subsetted in a way that reflects the extent to which users select the items of the search result when the items of the search result appear in search results generated for queries having one or more query terms in common with the received query.

4. The method of claim 1, wherein the search index contains both (1) explicit item fields for matching query terms and (2) implicit item fields indicating the extent to which users select the items of the generated search result when the items of the generated search result appear in search results generated for queries having one or more query terms in common with the received query and is used by the search engine to generate the generated search result.

5. The method of claim 1 wherein the items included in the generated query result with which are associated one or more of the query terms of the received query have textual content that matches one or more of the query terms of the received query.

6. The method of claim 1 wherein the items included in the generated query result with which are associated one or more of the query terms of the received query have textual metadata that matches one or more of the query terms of the received query.

7. The method of claim 1 wherein the generated search result contains items each of which is a different product.

8. The method of claim 1 wherein the generated search result contains items each of which is a different web page describing a product.

9. The method of claim 1 wherein the generated search result contains items each of which is a different web page.

10. The method of claim 1 wherein the generated search result contains items each of which is a different document.

11. The method of claim 1 wherein the generated search result further reflects the extent to which users fail to select the items of the generated search result when the items of the generated search result appear in search results generated for queries having one or more query terms in common with the received query.

12. A computer program product embedded in a computer-readable medium for maintaining an item index able to be used by any search engine able to utilize an item index to generate a search result, the computer program product including instructions that, when executed by a processor, cause the processor to perform steps comprising:

determining that, in a distinguished search result produced by the search engine in response to a distinguished query containing a distinguished query term, a user has selected a distinguished item from the distinguished search result; and

in response to so determining, updating in an index entry for the distinguished item the contents of an item selection field to reflect a tendency of users to select the distinguished item from search results produced by the search engine in response to queries containing the distinguished query term, the contents of the item selection field including one or more query terms each repeated a number of times representing a level of relevance of each of the one or more query terms to the item, the level of relevance corresponding to the tendency of users to select the distinguished item, the updating including adjusting the number of times at least one of the one or more query terms is repeated in the item selection field,

wherein the search engine is able to generate the search results based on the item index including the item selection field by determining an ordered set of items relating to a received query, an order of the ordered set of items being determined at least in part by generating a ranking score for each item in the ordered set of items, the ranking score being generated for each item in the ordered set of items by assigning a field score for each occurrence of one of the one or more query terms in the item selection field and summing the field scores for each item, the search results being generated without modifying operation of the search engine.

13. The computer-readable medium of claim 12 wherein the updating comprises adding the distinguished query term to the contents of an item selection field contained by the index entry for the distinguished item.

14. The computer-readable medium of claim 13 wherein the item selection field contained by the index entry for the distinguished item contains, for each of a plurality of terms, a number of occurrences of the term that reflects the frequency with which users select the distinguished item from search results produced by the search engine in response to queries containing the term.

15. The computer-readable medium of claim 12 wherein it is determined that the user has selected the distinguished item from the distinguished search result by communicating an instruction to purchase the distinguished item.

16. The computer-readable medium of claim 12 wherein it is determined that the user has selected the distinguished item from the distinguished search result by performing at least a threshold level of interaction with the item.

17. The computer-readable medium of claim 12 wherein it is determined that the user has selected the distinguished item from the distinguished search result by adding the distinguished item to a shopping basket.

18. The computer-readable medium of claim 12 wherein it is determined that the user has selected the distinguished item from the distinguished search result by viewing information relating to the distinguished item for at least a threshold period of time.

19. One or more computer memories collectively storing a search engine index data structure and, for each of a plurality of items indexed by the search engine index data structure, comprising:

at least one substantive field containing query terms that are among data or metadata of the item; and

at least one item selection field containing one or more query terms, each of the one or more query terms contained in the at least one item selection field being a query term that, when included in queries whose search results contain the item, frequently coincide with a user selecting the item from such search results, the one or more query terms in the at least one item selection field being repeated a number of times representing a level of relevance of each of the one or more query terms to the item, the level of relevance based at least in part upon a number of users having selected the item from the search results,

the contents of the search engine data structure being usable by any search engine, able to use the index data structure to generate search results, to determine an ordered set of items relating to a received query, an order of the ordered set of items being determined at least in part by generating a ranking score for each item in the ordered set of items, the ranking score being generated for each item in the ordered set of items by assigning a field score for each occurrence of one of the one or more query terms in the at least one item selection field and the at least one substantive field and summing the field scores for each item,

wherein the search engine is able to generate the search results without modifying operation of the search engine.

20. The computer memories of claim 19 wherein the item selection field contains a number of occurrences of each term that is directly related to the relative rate of selection for each of the terms.

21. A computing system for processing a query made up of one or more query terms, comprising:

a processor and memory collectively providing,

a search engine capable of processing a received query against an index, the search engine being any search engine able to utilize an index to generate search results; and

an index that, when used by the search engine to process the received query, causes the search engine to produce a query result that both includes items having data or metadata matching the query terms of the received query and reflects an extent to which users have selected the items of the generated search result when the items of the generated search result have appeared in search results generated for previous queries having one or more query terms in common with the received query, the index including at least one item selection field containing one or more query terms each repeated a number of times representing a level of relevance of the each of the one or more query terms to the item the level of relevance based at least in part upon the extent to which users have selected the item from the generated search result,

wherein the search engine produces the query result by determining an ordered set of items relating to the received query, an order of the ordered set of items being determined at least in part by generating a ranking score for each item in the ordered set of items, the ranking score being generated for each item in the ordered set of items by assigning a field score for each occurrence of one of the one or more query terms in the at least one item selection field and summing the field scores for each item,

wherein the search engine is able to generate the search results without modifying operation of the search engine.

22. The computing system of claim 21 wherein the index indicates, for each of a plurality of items, both terms that are among data or metadata for the item, as well as terms that have commonly occurred in queries from whose search results the item has been selected.

23. The computing system of claim 21, further comprising an index augmentation subsystem that periodically augments the index to reflect recent item selection information.