Freshness based ranking

• US 9,189,526 B1
• Filed: 03/15/2013
• Issued: 11/17/2015
• Est. Priority Date: 03/21/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Chen, Zhihui, Frankle, Jonathan
• Primary Examiner(s): Corrielus, Jean M
• Application Number:
  US13/832,208
• Time in Patent Office: 977 Days
• Field of Search: 707/737, 707/725, 707/749, 707/727, 707/723
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results

Abstract

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for ranking search results. One of the methods includes receiving a search result obtained in response to a query, wherein the search result identifies a resource and has an associated score S. A determination is made that users prefer newer resources over older resources for the query and that the resource is a new resource. A new score S′ is associated with the resource in place of S, based on the determination that users prefer newer resources over older resources for the query and that the resource is a new resource.

Claims

1. A computer-implemented method comprising:

receiving a search result obtained in response to a query, wherein the search result identifies a resource and has an associated score S;

determining a query value Q for the query, the query value Q being based on a maximum of one or more query freshness signals, the query value Q representing whether users prefer newer resources over older resources for the query;

determining a resource value D for the resource, the resource value D being based on a function of an age of the resource F(age), wherein F(age) represents whether the resource is a new resource;

calculating a new score S′

as a product of S and a boosting function, the boosting function being a function of the query value Q and the resource value D, wherein S′

>

S whenever the query value Q indicates that users prefer newer resources over older resources for the query and F(age) indicates that the resource is a new resource; and

associating the new score S′

with the search result in place of S.

2. The method of claim 1, wherein the boosting function is a continuous function having a range of values that includes values less than 1, a value equal to 1, and values greater than 1, whereby the product of S and the boosting function can be less than S, equal to S, or greater than S.

3. The method of claim 1, wherein the boosting function has the form Q^D.

4. The method of claim 1, wherein determining Q comprises determining a maximum of one or more of the following query freshness signals QFval:

a percentile value of a number of occurrences of the query being received within a first recent time period;

a percentile value of a number of occurrences of the query on blog web pages within a second recent time period;

a percentile value of a number of occurrences of the query on news web pages within a third recent time period;

a percentile value of a number of occurrences of the query on social network web pages within a fourth recent time period;

a percentile value of a number of occurrences of the query requesting news search results within a fifth recent time period;

a percentile value of a ratio of a number of occurrences of the query requesting news search results within a sixth recent time period to a number of occurrences of the query requesting web search results within the sixth recent time period;

a percentile value of a number of user selections of news search results provided in response to the query within a seventh recent time period; and

a percentile value of a ratio of a number of user selections of news search results provided in response to the query to a number of user selections of web search results provided in response to the query within an eighth recent time period.

5. The method of claim 4, wherein Q satisfies:

Q=Max(QFval)+1.

6. The method of claim 1, wherein F(age) is a monotonically decreasing function of an age of the resource.

7. The method of claim 6, wherein F(age) is a reverse sigmoid function.

8. The method of claim 1, wherein D is based on F(age) adjusted by a quality of the resource.

9. The method of claim 8, wherein the quality of the resource is based on one or more of the following functions:

a query-independent quality score G(provider) for a provider of the resource,

a query-dependent quality score H(query, provider) for a provider of the resource, and

a measure I(query, resource) of how well the query matches the resource.

10. The method of claim 9, wherein H(query, provider) represents a likelihood that users that issue the query will select a resource from a provider that created the resource.

11. The method of claim 10, wherein a provider is a user account associated with the resource.

12. The method of claim 9, wherein D satisfies:

13. A system comprising:

one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising;

receiving a search result obtained in response to a query, wherein the search result identifies a resource and has an associated score S;

determining a query value Q for the query, the query value Q being based on a maximum of one or more query freshness signals, the query value Q representing whether users prefer newer resources over older resources for the query;

determining a resource value D for the resource, the resource value D being based on a function of an age of the resource F(age), wherein F(age) represents whether the resource is a new resource;

calculating a new score S′

as a product of S as a boosting function, the boosting function being a function of the query value Q and the resource value D, wherein S′

>

S whenever the query value Q indicates that users prefer newer resources over older resources for the query and F(age) indicates that the resource is a new resource; and

associating the new score S′

with the search result in place of S.

14. The system of claim 13, wherein the boosting function is a continuous function having a range of values that includes values less than 1, a value equal to 1, and values greater than 1, whereby the product of S and the boosting function can be less than S, equal to S, or greater than S.

15. The system of claim 13, wherein the boosting function has the form Q^D.

16. The system of claim 13, wherein determining Q comprises determining a maximum of one or more of the following query freshness signals QFval:

a percentile value of a number of occurrences of the query being received within a first recent time period;

a percentile value of a number of occurrences of the query on blog web pages within a second recent time period;

a percentile value of a number of occurrences of the query on news web pages within a third recent time period;

a percentile value of a number of occurrences of the query on social network web pages within a fourth recent time period;

a percentile value of a number of occurrences of the query requesting news search results within a fifth recent time period;

a percentile value of a ratio of a number of occurrences of the query requesting news search results within a sixth recent time period to a number of occurrences of the query requesting web search results within the sixth recent time period;

a percentile value of a number of user selections of news search results provided in response to the query within a seventh recent time period; and

a percentile value of a ratio of a number of user selections of news search results provided in response to the query to a number of user selections of web search results provided in response to the query within an eighth recent time period.

17. The system of claim 16, wherein Q satisfies:

Q=Max(QFval)+1.

18. The system of claim 13, wherein D is based on F(age) adjusted by a quality of the resource.

19. The system of claim 18, wherein the quality of the resource is based on one or more of the following functions:

a query-independent quality score G(provider) for a provider of the resource,

a query-dependent quality score H(query, provider) for a provider of the resource, and

a measure I(query, resource) of how well the query matches the resource.

20. The system of claim 19, wherein H(query, provider) represents a likelihood that users that issue the query will select a resource from a provider that created the resource.

21. The system of claim 13, wherein F(age) is a monotonically decreasing function of an age of the resource.

22. The system of claim 21, wherein F(age) is a reverse sigmoid function.

23. A computer program product, encoded on one or more non-transitory computer storage media, comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

receiving a search result obtained in response to a query, wherein the search result identifies a resource and has an associated score S;

determining a query value Q for the query, the query value Q being based on a maximum of one or more query freshness signals, the query value Q representing whether users prefer newer resources over older resources for the query;

determining a resource value D for the resource, the resource value D being based on a function of an age of the resource F(age), wherein F(age) represents whether the resource is a new resource;

calculating a new score S′

as a product of S as a boosting function, the boosting function being a function of the query value Q and the resource value D, wherein S′

>

S whenever the query value Q indicates that users prefer newer resources over older resources for the query and F(age) indicates that the resource is a new resource; and

associating the new score S′

with the search result in place of S.