Ranking documents based on user behavior and/or feature data

• US 10,152,520 B1
• Filed: 03/30/2016
• Issued: 12/11/2018
• Est. Priority Date: 06/17/2004
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google LLC (Alphabet Inc.)
• Inventors: Dean, Jeffrey A., Anderson, Corin, Battle, Alexis
• Primary Examiner(s): Lin, Shew Fen
• Application Number:
  US15/085,826
• Time in Patent Office: 986 Days
• CPC Class Codes: G06F 16/24578 using ranking
• 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
• CPC Class Codes: G06F 40/134 Hyperlinking

Abstract

A system generates a model based on feature data relating to different features of a link from a linking document to a linked document and user behavior data relating to navigational actions associated with the link. The system also assigns a rank to a document based on the model.

Claims

1. A method comprising:

generating, by one or more server devices, a measure indicating a probability that links associated with particular feature data will be selected, the particular feature data identifying one or more link features associated with the links, wherein generating the measure comprises;

identifying link selection data for a plurality of documents collectively having corresponding links, of the links associated with the particular feature data, wherein each of the plurality of documents each has a corresponding link of the corresponding links, wherein the corresponding links collectively link to a plurality of target documents and each of the corresponding links is to a corresponding one of the plurality of target documents, the link selection data indicating;

a quantity of selections of the corresponding links of the plurality of documents, the selections each being based on a respective one of a plurality of user devices accessing a respective one of the documents and selecting the corresponding link of the respective one of the documents of the plurality of documents, and

generating, based on the link selection data, the measure indicating the probability that links associated with the particular feature data will be selected;

after generating the measure;

generating, by one or more of the server devices, a rank for a particular document, wherein generating the rank comprises;

determining that a particular link to the particular document is associated with the particular feature data,

determining a weight of the particular link based on the measure, wherein determining the weight based on the measure is based on determining that the particular link is associated with the particular feature data, and

generating the rank for the particular document based on the weight for the particular link, and based on additional weights for additional links to the particular document;

identifying, by one or more of the server devices, documents associated with a search query, the documents, associated with the search query, including the particular document; and

providing, by the one or more server devices, information associated with the particular document based on;

the search query, and

the generated rank for the particular document.

2. The method of claim 1, further comprising:

identifying, from memory associated with the one or more server devices, additional link selection data for a plurality of additional documents each having a corresponding additional link associated with additional particular feature data,

the additional particular feature data identifying one or more additional link attributes of the corresponding additional links, and

the additional link selection data indicating an additional quantity of additional user device selections of the corresponding additional links of the documents;

wherein generating the measure comprises generating the measure based on the link selection data and the additional link selection data.

3. The method of claim 2, wherein the additional particular feature data identifies one or more additional link features of the corresponding additional links.

4. The method of claim 2, wherein the particular feature data identifies the one or more link features associated with the links and identifies features of source documents containing the links or features of the target documents referenced by the links.

5. The method of claim 1, wherein generating the rank for the particular document further comprises:

determining that the particular link to the particular document is associated with additional particular feature data;

identifying an additional measure indicating an additional probability that links associated with the additional particular feature data will be selected; and

determining the weight of the particular link based on the measure and the additional measure, wherein determining the weight based on the measure and the additional measure is based on the particular link having the particular feature data and having the additional particular feature data.

6. The method of claim 1, wherein the one or more link features identify at least one of:

context relating to one or more words before or after the links;

words in anchor text associated with the links; and

a quantity of the words in the anchor text.

7. The method of claim 6, wherein the one or more link features identify each of:

context relating to one or more words before or after the links;

words in anchor text associated with the links; and

a quantity of the words in the anchor text.

8. The method of claim 1, wherein the particular feature data comprises features embodied by the links and identifies features of source documents containing the links or features of the target documents referenced by the links.

9. The method of claim 8, wherein the particular feature data identifies the one or more link features, identifies features of the source documents containing the links, and identifies features of the target documents referenced by the links.

10. The method of claim 1, further comprising:

sorting information regarding the documents based on ranks associated with the documents; and

providing the sorted information regarding the documents,

where providing the sorted information regarding the documents includes providing the information associated with the particular document.

11. One or more devices comprising:

one or more memories to store instructions; and

one or more processors to execute the instructions to;

generate a measure indicating a probability that links associated with particular feature data will be selected, the particular feature data identifying one or more link features associated with the links, wherein when generating the measure, the one or more of the processors are to;

identify link selection data for a plurality of documents collectively having corresponding links, of the links associated with the particular feature data, wherein each of the plurality of documents each has a corresponding link of the corresponding links, wherein the corresponding links collectively link to a plurality of target documents and each of the corresponding links is to a corresponding one of the plurality of target documents, the link selection data indicating;

a quantity of selections of the corresponding links of the plurality of documents, the selections each being based on a respective one of a plurality of user devices accessing a respective one of the documents and selecting the corresponding link of the respective one of the documents of the plurality of documents, and

generating, based on the link selection data, the measure indicating the probability that links associated with the particular feature data will be selected;

after generating the measure;

generate a rank for a particular document, wherein when generating the rank, the one or more of the processors are to;

determine that a particular link to the particular document is associated with the particular feature data,

determine a weight of the particular link based on the measure, wherein determining the weight based on the measure is based on determining that the particular link is associated with the particular feature data, and

generate the rank for the particular document based on the weight for the particular link, and based on additional weights for additional links to the particular document;

identify documents associated with a search query, the documents, associated with the search query, including the particular document; and

provide information associated with the particular document based on;

the search query, and

the generated rank for the particular document.

12. The one or more devices of claim 11, wherein the one or more processors are further to:

identify, from memory associated with the one or more processors, additional link selection data for a plurality of additional documents each having a corresponding additional link associated with additional particular feature data,

the additional particular feature data identifying one or more additional link attributes of the corresponding additional links, and

the additional link selection data indicating an additional quantity of additional user device selections of the corresponding additional links of the documents;

wherein when generating the measure, the one or more processors are further to;

generate the measure based on the link selection data the additional link selection data.

13. The one or more devices of claim 12, wherein the additional particular feature data identifies one or more additional link features of the corresponding additional links.

14. The one or more devices of claim 12, wherein the particular feature data identifies the one or more link features associated with the links and identifies features of source documents containing the links or features of the target documents referenced by the links.

15. The one or more devices of claim 11, wherein when generating the rank for the particular document, the one or more processors are further to:

determine that the particular link to the particular document is associated with additional particular feature data;

identify an additional measure indicating an additional probability that links associated with the additional particular feature data will be selected; and

determine the weight of the particular link based on the measure and the additional measure, wherein determining the weight based on the measure and the additional measure is based on the particular link having the particular feature data and having the additional particular feature data.

16. The one or more devices of claim 11, wherein the one or more link features identify at least one of:

context relating to one or more words before or after the links;

words in anchor text associated with the links; and

a quantity of the words in the anchor text.

17. The one or more devices of claim 16, wherein the one or more link features identify each of:

the context relating to the one or more words before or after the links;

the words in the anchor text associated with the links; and

the quantity of the words in the anchor text.

18. The one or more devices of claim 11, wherein the particular feature data comprises features of the links and identifies features of source documents containing the links or features of the target documents referenced by the links.

19. The one or more devices of claim 11, wherein the one or more processors are further to:

sort information regarding the documents based on ranks associated with the documents; and

provide the sorted information regarding the documents,

where providing the sorted information regarding the documents includes providing the information associated with the particular document.

20. A non-transitory computer-readable medium storing instructions, the instructions comprising:

a plurality of instructions which, when executed by one or more devices, cause the one or more devices to;

generate a measure indicating a probability that links associated with particular feature data will be selected, the particular feature data identifying one or more link features associated with the links, wherein when generating the measure, the one or more devices are to;

identify link selection data for a plurality of documents collectively having corresponding links, of the links associated with the particular feature data, wherein each of the plurality of documents each has a corresponding link of the corresponding links, wherein the corresponding links collectively link to a plurality of target documents and each of the corresponding links is to a corresponding one of the plurality of target documents, the link selection data indicating;

a quantity of selections of the corresponding links of the plurality of documents, the selections each being based on a respective one of a plurality of user devices accessing a respective one of the documents and selecting the corresponding link of the respective one of the documents of the plurality of documents, and

generate, based on the link selection data, the measure indicating the probability that links associated with the particular feature data will be selected;

after generating the measure;

generate a rank for a particular document, wherein when generating the rank one or more of the devices are to;

determine that a particular link to the particular document is associated with the particular feature data,

determine a weight of the particular link based on the measure, wherein determining the weight based on the measure is based on determining that the particular link is associated with the particular feature data, and

generate the rank for the particular document based on the weight for the particular link, and based on additional weights for additional links to the particular document;

identify documents associated with a search query, the documents, associated with the search query, including the particular document; and

provide information associated with the particular document based on;

the search query, and

the generated rank for the particular document.