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

• US 6,185,558 B1
• Filed: 03/10/1998
• Issued: 02/06/2001
• Est. Priority Date: 03/03/1998
• Status: Expired due to Term

• Current Assignee: A9.com Incorporated (Amazon.com, Inc.)
• Original Assignee: Amazon.com, Inc.
• Inventors: Linden, Greg, Bowman, Dwayne, Ortega, Ruben E., Spiegel, Joel R.
• Primary Examiner(s): Alam, Hosain T.
• Application Number:
  US09/041,081
• Time in Patent Office: 1,064 Days
• Field of Search: 707/1-6, 707/10, 707/501, 707/513, 709/200, 709/201, 709/217-219, 705/35-37, 705/7, 705/10
• US Class Current: N/A
• CPC Class Codes: G06F 16/3325 Reformulation based on results of preceding query
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/3349 Reuse of stored results of previous queries
• CPC Class Codes: G06F 16/338 Presentation of query results
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 40/04 Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

The present invention provides a software facility for identifying the items most relevant to a current query based on items selected in connection with similar queries. In preferred embodiments of the invention, the facility receives a query specifying one or more query terms. In response, the facility generates a query result identifying a plurality of items that satisfy the query. The facility then produces a ranking value for at least a portion of the items identified in the query result by combining the relative frequencies with which users selected that item from the query results generated from queries specifying each of the terms specified by the query. The facility identifies as most relevant those items having the highest ranking values.

Claims

1. A method in a computer system for ranking items in a search result, the method comprising:

receiving a rating set of queries, each query in the rating set specifying one or more terms;

for each query in the rating set,

generating a query result identifying one or more items satisfying the query;

allowing a user to select one or more of the items identified in the query result;

for each item selected from the query result, for each term specified by the query, increasing a rating value corresponding to the combination of the selected item and the term specified by the query, the rating value indicating a relative frequency with which users have selected the selected item when the selected item has been identified in query results generated from queries containing the term specified by the query;

receiving a distinguished query specifying one or more terms;

generating a distinguished query result identifying a plurality of items satisfying the distinguished query; and

for each item identified in the distinguished query result,

for the rating values corresponding to the combination of the item identified in the distinguished query result and one of the terms specified by the distinguished query, combining these rating values to generate a ranking value for the item within the distinguished query result.

2. The method of claim 1, further comprising imposing on the items identified in the distinguished query result an order in which the ranking values of the items monotonically decreases.

3. The method of claim 1, further comprising using the ranking values of the items identified in the distinguished query result to create a proper subset of the items.

4. The method of claim 3 wherein the creating creates a subset of the items identified in the distinguished query result that contains all of the items whose ranking values exceed a predetermined minimum ranking value.

5. The method of claim 3 wherein the creating creates a subset of the items identified in the distinguished query result that contains all of the items whose ranking values exceed a predetermined minimum ranking value.

6. The method of claim 1 wherein the increasing increases rating values for selections made to display additional information about items.

7. The method of claim 1 wherein the increasing increases rating values for selections made to purchase items.

8. The method of claim 1 wherein the increasing increases rating values for selections made to add items to a tentative list of purchases.

9. The method of claim 1 wherein the increasing increases rating values for selections of portions of detailed information displayed about items.

10. The method of claim 1 wherein the increasing increases rating values for units of time for which the user displays detailed information about items.

11. A method in a computer system for ranking items in a search result, the method comprising:

for each of a multiplicity of search terms, compiling data indicating the extent to which users have selected each of a multiplicity of items when returned in search results produced from queries containing the search term;

receiving a query and a search result, the received query containing a term among the multiplicity of terms, the received search result identifying a plurality of items among the multiplicity of items that satisfy the received query; and

using the compiled data to rank at least a portion of the items identified in the received search result in accordance with the extent to which users have selected each of the plurality of items identified in the received search result when returned in search results produced from queries containing the search term contained in the received query.

12. The method of claim 11 wherein at least a portion of the users are identified with one of a plurality of demographic groups, and wherein the compiling step compiles, for each demographic group of the plurality of demographic groups, data indicating the extent to which users identified with the demographic group have selected each of a multiplicity of items when returned in search results produced from queries containing the search term, and wherein the received query is submitted on behalf of a distinguished user identified with a distinguished demographic group, and wherein the ranking step uses the compiled data to rank the items identified in the received search result in accordance with the extent to which users identified with the distinguished demographic group have selected each of the plurality of items identified in the received search result when returned in search results produced from queries containing the search term contained in the received query.

13. The method of claim 11, further comprising imposing on the items identified in the distinguished search result an order in which the ranking values of the items monotonically decreases.

14. The method of claim 11, further comprising using the ranking values of the items identified in the distinguished search result to create a proper subset of the items.

15. The method of claim 14 wherein the creating creates a subset of the items identified in the distinguished search result that contains all of the items whose ranking values exceed a predetermined minimum ranking value.

16. The method of claim 11, further comprising improving the ranks of items for selections made to display additional information about items.

17. The method of claim 11, further comprising improving the ranks of items for selections made to purchase items.

18. The method of claim 11, further comprising improving the ranks of items for selections made to add items to a tentative list of purchases.

19. The method of claim 11, further comprising improving the ranks of items for selections of portions of detailed information displayed about items.

20. The method of claim 11 wherein the increasing increases rating values for units of time for which the user displays detailed information about items.

21. A computer-readable medium whose contents cause a computer system to rank items in a search result by:

for each of a multiplicity of search terms, compiling data indicating the extent to which users have selected each of a multiplicity of items when returned in search results produced from queries containing the search term;

receiving a query and a search result, the received query containing a term among the multiplicity of terms, the received search result identifying a plurality of items among the multiplicity of items that satisfy the received query; and

using the compiled data to rank at least a portion of the items identified in the received search result in accordance with the extent to which users have selected each of the plurality of items identified in the received search result when returned in search results produced from queries containing the search term contained in the received query.

22. A method in a computer system for ranking items in a search result, the method comprising:

receiving a query specifying one or more terms;

generating a query result identifying a plurality of items satisfying the query; and

for a plurality of items identified in the query result, combining ratings of frequencies with which users selected the item in earlier queries specifying one or more terms of the query to producing a ranking value for the item.

23. The method of claim 22, further comprising adjusting the produced ranking values to reflect the number of terms specified by the query that are matched by the item.

24. The method of claim 22, further comprising imposing on the items identified in the distinguished query result an order in which the ranking values of the items monotonically decreases.

25. The method of claim 22, further comprising using the ranking values of the items identified in the distinguished query result to create a proper subset of the items.

26. The method of claim 25 wherein the creating creates a subset of the items identified in the distinguished query result that contains all of the items whose ranking values exceed a predetermined minimum ranking value.

27. The method of claim 25 wherein the creating creates a subset of the items identified in the distinguished query result that contains all of the items whose ranking values exceed a predetermined minimum ranking value.

28. A computer-readable medium whose contents cause a computer system to rank items in a search result by:

receiving a query specifying one or more terms;

generating a query result identifying a plurality of items satisfying the query; and

for each item identified in the query result, combining the relative frequencies with which users selected the item in earlier queries specifying each of the terms of the query to producing a ranking value for the item.

29. The computer-readable medium of claim 28 wherein the contents of the computer-readable medium further cause the computer system to perform the step of adjusting the ranking value produced for each item identified in the query result to reflect the number of terms specified by the query that are matched by the item.

30. A method in a computer system for identifying significant items satisfying a query, the method comprising:

receiving a query specifying a plurality of terms;

for each term specified by the query, identifying a predetermined number of items that users have selected most frequently in earlier queries specifying the term;

for each unique item identified in the identifying step, producing a ranking value for the item by combining one or more indications of the relative frequencies with which users have selected the item in earlier queries specifying one of the terms specified by the received query; and

selecting as the significant items a predetermined number of items having the largest ranking values.

31. The method of claim 30, further comprising displaying information relating to the selected items in a manner distinct from that in which information about unselected items is displayed.

32. A computer-readable medium whose contents cause a computer system to identify the most significant items satisfying a query by:

receiving a query specifying one or more terms;

for each term specified by the query, identifying a predetermined number of items having the largest relative frequencies with which users selected the items in earlier queries specifying the term;

for each unique item identified in the identifying step, combining the relative frequencies for the item to produce a ranking value for the item; and

selecting as the most significant items a predetermined number of items having the largest ranking values.

33. A computer system for ranking items in a search result, comprising:

a query memory that stores information about previously submitted queries and items selected from the query results of previously submitted queries;

a query receiver that receives queries each specifying one or more terms;

a query server that generating a query result for each query received by the query receiver that identifies a plurality of items satisfying the query; and

an item ranking subsystem that, for each query result generated by the query server, for at least a portion of the items identified in the query result, combines from the contents of the query memory the relative frequencies with which users selected the item in earlier queries specifying each of the terms of the query to producing a ranking value for the item.

34. A method in a computer system for compiling statistics usable to rank items in a distinguished query result produced for a distinguished query, the method comprising:

receiving a rating set of queries, each query in the rating set specifying one or more terms;

for each query in the rating set,

generating a query result identifying one or more items satisfying the query;

allowing a user to select one or more of the items identified in the query result; and

for items selected from the query result, for terms specified by the query, adjusting a rating score corresponding to the combination of the selected item and the term specified by the query, the rating score indicating the relative frequency with which users have selected the selected item when the selected item has been identified in search results generated from queries containing the search term specified by the query, to produce rating scores usable to rank items in a distinguished query result produced for a distinguished query.

35. The method of claim 34 wherein the adjusting is performed in real-time with user selections.

36. The method of claim 34 wherein the adjusting is performed periodically for a batch of earlier-occurring user selections.

37. The method of claim 36 wherein the queries received are each identified with a different time period, and wherein the adjusting for each query increases a rating score corresponding to the time period with which the query is identified, to produce a different body of rating scores for each time period, and wherein the method further comprises combining a subset of the produced bodies of rating scores to generate a composite body of rating scores usable to order items in a distinguished query result produced for a distinguished query.

38. The method of claim 37 wherein the combining weights scores from different bodies of rating scores differently depending on the recency of the time periods for which the bodies of rating scores are produced.

39. The method of claim 34 wherein each query term is specified for one of a plurality of query fields, and wherein the adjusting step for each term increases a rating score corresponding to the query field for which the term is identified, to produce different rating scores for each different field for which a query term is specified.

40. The method of claim 34, further comprising:

for each received query, generating a query log entry, the query log entry containing the terms specified by the query and a query identifier identifying the query;

for each user selection of an item identified in a query result generated for a received query, generating a selection log entry, the selection log entry containing a query identifier identifying the query and an item identifier identifying the item selected; and

identifying each user selection by;

identifying a selection log entry;

extracting from the identified selection log entry the query identifier and the item identifier;

identifying a query log entry containing the extracted query identifier; and

extracting from the identified query log entry the terms specified by the query.

41. The method of claim 36, further comprising:

for each user selection of an item identified in a query result generated for a received query, generating a database record identifying both the terms of the query and the item selected; and

identifying each user selection by;

retrieving a database record, and

extracting from the retrieved database record the terms of the query and the item selected.

42. The method of claim 34 wherein the adjusting increases rating values for selections made to display additional information about items.

43. The method of claim 34 wherein the adjusting increases rating values for selections made to purchase items.

44. The method of claim 34 wherein the adjusting increases rating values for selections made to add items to a tentative list of purchases.

45. The method of claim 34 wherein the adjusting increases rating values for selections of portions of detailed information displayed about items.

46. The method of claim 34 wherein the adjusting increases rating values for units of time for which the user displays detailed information about items.

47. A computer-readable medium whose contents cause a computer system to compile statistics usable to rank items in a distinguished query result produced for a distinguished query by:

receiving a rating set of queries, each query in the rating set specifying one or more terms;

for each query in the rating set,

generating a query result identifying one or more items satisfying the query;

allowing a user to select one or more of the items identified in the query result; and

for each item selected from the query result, for each term specified by the query, increasing a rating score corresponding to the combination of the selected item and the term specified by the query, the rating score indicating the relative frequency with which users have selected the selected item when the selected item has been identified in search results generated from queries containing the search term specified by the query, to produce a body of rating scores usable to rank items in a distinguished query result produced for a distinguished query.

48. The computer-readable medium of claim 47 wherein the queries received in the receiving step are each identified with a different time period, and wherein the increasing step for each query increases a rating score corresponding to the time period with which the query is identified, to produce a different body of rating scores for each time period, and wherein the computer readable medium further causes the computer system to perform the step of combining a subset of the produced bodies of rating scores to generate a composite body of rating scores usable to order items in a distinguished query result produced for a distinguished query.

49. A computer memory containing a user behavior data structure usable to rank the relevance of items in a query result, the data structure comprising a plurality of rating scores, each rating score corresponding both to a query term and to an item, and reflecting quantitatively the extent to which users have selected the item from query results generated from queries specifying the query term, such that the data structure may be used to rank items in a distinguished query result produced for a distinguished query by, for each item in the distinguished query result, retrieving from the data structure the rating scores corresponding to the item and any term specified in the distinguished query and combining the retrieved rating scores to generate a ranking value for the item.

50. A method in a computer system for handling a query, comprising:

receiving a query containing one or more terms;

processing the query to produce a result set of items that satisfy the query;

where the result set contains no items, identifying one or more items that, though they do not satisfy the received query, have been selected by users relatively frequently when presented in response to queries containing at least some of the same terms as the received query; and

displaying the identified items,

such that items are displayed even though the result set contains no items.

51. The method of claim 50 wherein the identifying step comprises:

accessing rating scores indicating, for different combinations of terms and items, the frequency with which users have selected the item when presented in response to queries containing the term;

for each term contained in the query, gathering one or more items whose combination with the term have the largest rating scores among the rating scores for any item in combination with the term; and

identifying at least a portion of the gathered items.

52. A computer-readable medium whose contents cause a computer system to handle a query by:

receiving a query containing one or more terms;

processing the query to produce a result set of items that satisfy the query;

where the result set contains no items, identifying one or more items that, though they do not satisfy the received query, have been selected by users relatively frequently when presented in response to queries containing at least some of the same terms as the received query.

53. The computer-readable medium of claim 52 wherein the identifying includes:

accessing rating scores indicating, for different combinations of terms and items, the frequency with which users have selected the item when presented in response to queries containing the term;

for each term contained in the query, gathering one or more items whose combination with the term have the largest rating scores among the rating scores for any item in combination with the term; and

identifying at least a portion of the gathered items.