Rank-adjusted content items

• US 7,610,282 B1
• Filed: 03/30/2007
• Issued: 10/27/2009
• Est. Priority Date: 03/30/2007
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Datar, Mayur, Garg, Ashutosh, Dhamdhere, Kedar
• Primary Examiner(s): Le; Debbie M
• Application Number:
  US11/694,268
• Time in Patent Office: 942 Days
• Field of Search: 707 4- 5
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24575 using context
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

Click logs and query logs are processed to identify statistical search patterns. A search session is compared to the statistical search patterns. Content items responsive to a query of the search session are identified, and a ranking of the content items is adjusted based on the comparison.

Claims

1. A computer-implemented method, comprising:

selecting, by a computer, query paths, each query path being a plurality of queries sequenced in an order in which the queries were provided for a previous search session;

identifying, by the computer, content terminuses, each content terminus defining a terminus of a query path and corresponding to a selection of a content item in response to a query in the query path;

identifying, by the computer, a context of a current search session, wherein the context is defined by a plurality of search session queries of the current search session;

determining, by the computer, that the context is related to a selected query path of the selected query paths only if two or more of the queries of the selected query path are similar to two or more of the search session queries of the current search session;

identifying, by the computer, two or more content items responsive to a query of the current search session based on the determination, the two or more content items being associated with a content terminus;

for each of the two or more content items, determining a corresponding likelihood that the content item will be selected given the content terminus and respective query path; and

adjusting a ranking of each of the two or more content items in substantially proportional relation to the corresponding likelihood.

2. The method of claim 1, wherein:

identifying content terminuses comprises;

identifying content terminuses defined by click logs identifying a selection of a content item.

3. The method of claim 1, wherein:

determining that the context is related to a selected query path comprises;

identifying a union of the two or more of the queries of the selected query path and the two or more of the search session queries of the current search session.

4. The method of claim 3, comprising:

determining that the context is related to the selected query path if the identified union is an exclusive union.

5. A computer-implemented method, comprising:

analyzing, by a computer, click logs and query logs;

selecting, by the computer, statistical search patterns associated with content items based on the analysis of the click logs and query logs, wherein each statistical search pattern is defined by a plurality of queries sequenced in an order in which the queries were provided for a previous search session;

comparing, by the computer, a plurality of queries received for a current search session to the selected statistical search patterns;

determining, by the computer, that the current search session is related to a selected statistical search pattern only if two or more of the queries of the selected statistical search pattern are similar to two or more of the queries of the current search session;

identifying, by the computer, two or more content items responsive to a query of the current search session based on the determination, the two or more content items being associated with a statistical search pattern;

for each of the two or more content items, determining a corresponding likelihood that the content item will be selected based on the statistical search pattern; and

adjusting, by the computer, a ranking of each of the two or more content items in substantially proportional relation to the corresponding likelihood.

6. The method of claim 5, wherein:

the click logs define content items responsive to previous search session queries.

7. The method of claim 6, wherein:

the query logs comprise previous search session queries.

8. The method of claim 7, wherein selecting statistical search patterns comprises:

identifying content terminuses defined by the click logs; and

identifying query paths terminating at the content terminuses.

9. The method of claim 8, further comprising:

identifying a search session category for each of the query paths.

10. The method of claim 9, further comprising:

identifying a current search session category.

11. The method claim 10, further comprising:

determining whether the current search session category relates to one of the identified search session categories.

12. The method of claim 10, further comprising:

identifying a content category for each of the content terminuses; and

determining whether the current search session category relates to an identified content category.

13. The method of claim 5, wherein the content items comprise webpages.

14. A system, comprising:

a data processor;

a computer memory device;

mining engine stored in the computer memory device and upon execution by the data processor mines click logs and query logs and selects query paths and content terminuses associated with the query paths, each query path being a plurality of queries sequenced in an order in which the queries were provided in a previous search session, and each content terminus being a terminus of a query path and corresponding to a selection of a content item in response to a query in the query path; and

a contextual search engine stored in the computer memory device and upon execution by the data processor;

identifies a context in a current search session, the context defined by a plurality of queries of the current search session;

determines that the context is related to a content terminus only if two or more of the queries of the query path associated with the content terminus are similar to two or more of the queries of the current search session;

identifies two or more content items responsive to a query of the current search session based on the determination, the two or more content items being associated with a content terminus;

for each of the two or more content items, determine a corresponding likelihood that the content item will be selected given the content terminus and respective query path; and

adjusts a ranking of each of the two or more content items in substantially proportional relation to the corresponding likelihood.

15. A system, comprising:

means for selecting content terminuses associated with content items, each content terminus defining a terminus of a query path, and each query path being a plurality of queries sequenced in an order in which the queries were provided in a previous search session and each content terminus corresponding to a selection of a content item in response to a query in the query path;

means for identifying a context in a current search session, wherein the context is defined by a plurality of queries of the current search session;

means for determining that the context is related to a content terminus only if two or more of the queries of the query path associated with the content terminus are similar to two or more queries of the current search session; and

means for identifying two or more content items responsive to a query of the current search session based on the determination, the two or more content items being associated with a content terminus;

means for determining for each of the two or more content items a corresponding likelihood that the content item will be selected given the content terminus and respective query path; and

means for adjusting a ranking of each of the two or more content items in substantially proportional relation to the corresponding likelihood.