Identifying topical entities

• US 10,068,022 B2
• Filed: 06/03/2011
• Issued: 09/04/2018
• Est. Priority Date: 06/03/2011
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google LLC (Alphabet Inc.)
• Inventors: Pilpel, Haran, Shmiel, Tomer, Ofek, Eran, Barkai, Eldad, Bar-Yossef, Ziv
• Primary Examiner(s): Featherstone, Mark D
• Assistant Examiner(s): Gmahl, Navneet
• Application Number:
  US13/153,365
• Publication Number:
  US 20150278366A1
• Time in Patent Office: 2,650 Days
• Field of Search: 707720-725
• CPC Class Codes: G06F 16/9024 Graphs; Linked lists G06F16/9027 takes precedence
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying topical entities. In one aspect, a method includes obtaining a plurality of entities that are associated with a first resource; for one or more of the identified entities, receiving search results for a search query derived from the entity; determining that search results for a search query including a particular entity include a specific type of search results; and determining that the particular entity is a topical entity of the first resource based at least in part on the particular entity appearing in a title or a resource locator of the first resource, wherein the topical entity of the first resource represents a predominant topic of the first resource.

Claims

1. A computer-implemented method comprising:

identifying a plurality of entities that are associated with a particular web resource, each entity being a word or a phrase; and

determining whether one entity from the plurality of entities is a topical entity that represents a predominant topic of the resource, wherein the determining comprises;

generating respective search queries for each of the identified entities;

obtaining, from a search engine, respective search results responsive to each of the generated search queries, wherein each search result has a respective ranking score;

determining that a search result referencing the particular web resource appears above a specified rank in a ranking of the search results responsive to at least one generated search query for at least one entity; and

in response to determining that a search result referencing the particular web resource appears above a specified rank in the ranking of the search results responsive to at least one generated search query for at least one entity;

if the particular web resource appears above a specified rank in the ranking of the search results responsive to the generated search query for exactly one entity, designating the one entity as the topical entity; and

if a search result referencing the particular web resource appears above a specified rank in the ranking of multiple search results responsive to multiple generated search queries, each search query and responsive results associated with more than one entity, designating one entity that generated a search result with a highest ranking score, among the multiple search results referencing the particular web resource, as the topical entity for the particular web resource.

2. The method of claim 1, further comprising:

associating additional content with the particular web resource.

3. The method of claim 2, wherein associating additional content with the particular web resource comprises:

increasing ranking scores of one or more highest scoring search results that were responsive to the search query generated from the topical entity;

selecting at least one web resource referenced by a particular search result having an increased ranking score that was responsive to the search query generated from the topical entity; and

associating content from the selected web resource with the particular web resource for the presentation to a user who subsequently requests the particular web resource.

4. A computer-implemented method comprising:

identifying a topical entity for a particular web resource, comprising;

identifying multiple entities that are associated with the particular web resource, each entity being a word or a phrase;

generating respective search queries for each of the identified entities;

obtaining, from a search engine, respective search results responsive to each of the generated search queries, wherein each search result has a respective ranking score;

determining that, in multiple different search result sets responsive to respective different search queries generated from the entities associated with the particular web resource, at least one search result references the particular web resource within a threshold number of top-ranked search results, wherein the search result is responsive to a search query generated for a particular entity associated with the particular web resource;

in response to determining that, in multiple different search result sets responsive to respective search queries generated from the entities associated with the particular web resource, a single search result references the particular web resource within the threshold number of top-ranked search results, wherein the search result is responsive to a search query generated for a particular entity associated with the particular web resource, designating the particular entity as the topical entity for the particular web resource; and

in response to determining that, in multiple different search result sets responsive to respective search queries generated from the entities associated with the particular web resource, a respective search result references the particular web resource within the threshold number of top-ranked search results, wherein the respective search result is responsive to the respective one of multiple search queries, each generated for a respective entity associated with the particular web resource, designating one entity that generated a search result with a highest ranking score, among the multiple search results referencing the particular web resource, as the topical entity for the particular web resource.

5. The method of claim 4, further comprising:

receiving an indication that a user has requested the particular web resource; and

in response to the indication, providing the content from the selected one or more web resources associated with the particular web resource for presentation to the user along with the particular web resource.

6. The method of claim 4, further comprising:

associating additional content with the particular web resource.

7. The method of claim 6, wherein associating additional content with the particular web resource comprises:

increasing ranking scores of one or more highest scoring search results that were responsive to the search query generated from the topical entity;

selecting at least one web resource referenced by a particular search result having an increased ranking score that was responsive to the search query generated from the topical entity; and

associating content from the selected web resource with the particular web resource for the presentation to a user who subsequently requests the particular web resource.

8. A computer-implemented method for associating content comprising:

identifying a topical entity for a particular web resource, comprising;

identifying multiple entities that are associated with the particular web resource, each entity being a word or a phrase;

generating respective search queries for each of the identified entities,

obtaining, from a search engine, respective search results responsive to each of the generated search queries, wherein each search result has a respective ranking score,

determining that search results responsive to a search query generated from a particular entity include a particular type of search result;

determining that the particular entity occurs in a title or a resource locator of the particular web resource;

in response to determining that the search results responsive to the search query generated from the particular entity include the particular type of search result and that the particular entity occurs in the title or the resource locator of the particular web resource, designating the particular entity as a topical entity of the particular web resource; and

in response to determining that, in multiple different search result sets responsive to respective search queries generated from the entities associated with the particular web resource, multiple search results reference the particular web resource within the threshold number of top-ranked search results, wherein the web resource is responsive to multiple search queries, each generated for a particular entity associated with the particular web resource, designating one entity that generated a search result with a highest ranking score, among the multiple search results referencing the particular web resource, as the topical entity for the particular web resource.

9. The method of claim 8, wherein the particular type of search result is one of:

shopping results, map results, or an answer box.

10. The method of claim 8, further comprising:

receiving an indication that a user has requested a particular web resource; and

in response to the indication, providing the content from the selected one or more web resources associated with the particular web resource for presentation to the user along with the particular web resource.

11. The method of claim 8, further comprising:

associating additional content with the particular web resource.

12. The method of claim 11, wherein associating additional content with the particular web resource comprises:

increasing ranking scores of one or more highest scoring search results that were responsive to the search query generated from the topical entity;

selecting at least one web resource referenced by a particular search result having an increased ranking score that was responsive to the search query generated from the topical entity; and

associating content from the selected web resource with the particular web resource for the presentation to a user who subsequently requests the particular web resource.