Methods and apparatus for matching relevant content to user intention
First Claim
Patent Images
1. A method comprising:
- receiving a search query from a user;
identifying search results responsive to the search query;
determining a plurality of concepts matching the search query;
retrieving the plurality of concepts associated with the search results using a page-concept index;
determining related concepts related to the plurality of concepts matching the search query and the plurality of concepts associated with the search results using a concept graph, the concept graph being an automated taxonomy of online information resources, the concept graph comprising information on a plurality of concepts of the online information resources and further comprising relationships among the plurality of concepts;
generating a set of concepts, the set of concepts comprising the plurality of concepts matching the search query, the plurality of concepts associated with the search results and the related concepts;
clustering the set of concepts into a plurality of grouping of top concepts, wherein each grouping represents a potential user intention of the user;
identifying a potential user intention of the user corresponding to each grouping of top concepts, wherein each potential user intention comprises one aspect of the topic that the user could be looking for, when the user enters the search query;
matching each potential user intentions of the user to an advertisement; and
delivering each advertisement and the search results to the user.
6 Assignments
0 Petitions
Accused Products
Abstract
Methods and apparatus for a new approach to the problem of matching relevant content to user queries. Instead of looking for the exact keyword, the invention expands it into groupings of concepts and phrases, where each such group represents one possible user intention (as implied by the query phrase or keyword). Each such grouping is analyzed to provide relevant content, including but not limited to, unstructured data like world wide web, categorized data and paid listings. The provided method can better capture user intentions even for cases where there is no click-through information.
158 Citations
15 Claims
-
1. A method comprising:
-
receiving a search query from a user; identifying search results responsive to the search query; determining a plurality of concepts matching the search query; retrieving the plurality of concepts associated with the search results using a page-concept index; determining related concepts related to the plurality of concepts matching the search query and the plurality of concepts associated with the search results using a concept graph, the concept graph being an automated taxonomy of online information resources, the concept graph comprising information on a plurality of concepts of the online information resources and further comprising relationships among the plurality of concepts; generating a set of concepts, the set of concepts comprising the plurality of concepts matching the search query, the plurality of concepts associated with the search results and the related concepts; clustering the set of concepts into a plurality of grouping of top concepts, wherein each grouping represents a potential user intention of the user; identifying a potential user intention of the user corresponding to each grouping of top concepts, wherein each potential user intention comprises one aspect of the topic that the user could be looking for, when the user enters the search query; matching each potential user intentions of the user to an advertisement; and delivering each advertisement and the search results to the user. - View Dependent Claims (2, 3, 4, 5, 6, 7)
-
-
8. A computer system comprising:
-
memory; and a processor coupled to the memory, the processor comprising; an input information module to receive a search query; a page-concept index comprising an inverted index of a plurality of categorized data items, each data item being tagged with related concept information; a search engine to obtain top categorized data items from online information resources; an advertising information pool comprising advertising materials stored in the memory, the advertising materials tagged with at least one concept or at least one context; a concept graph stored in the memory, the concept graph being an automated taxonomy of the online information resources, the concept graph comprising information on a plurality of concepts and relationships among the plurality of concepts, wherein the concept graph is stored in the memory prior to the input information being received at the input information module; a query refinement engine to determine advertising material relevant to a user intention by determining a user intention of the search query based on the search query, the categorized data items, the related concept information, the plurality of concepts, and the relationships among the plurality of concepts, wherein the user intention is a plurality of groupings of the top related concepts, where each of the plurality of groupings covers one aspect of the topic that the user could be looking for when the user inputs the search query, and wherein determining the user intention of the search query comprises clustering the plurality of concepts and related concepts into a grouping of top concepts, and wherein the relationships among the plurality of concepts in the concept graph define a boundary of the grouping of top concepts, wherein each grouping represents a potential user intention of the user; and an output interface to output the advertising material from the advertising information pool, wherein the advertising material is identified by matching an advertisement to the user intention by matching the user intention with concept tags in the advertisement database. - View Dependent Claims (9, 10, 11)
-
-
12. A non-transitory computer-readable storage medium embodying a set of computer-readable instructions implementing a method comprising:
-
receiving a search query from a user; identifying search results responsive to the search query; determining a plurality of concepts matching the search query; retrieving the plurality of concepts associated with the search results using a page-concept index; determining related concepts related to the plurality of concepts matching the search query and the plurality of concepts associated with the search results using a concept graph, the concept graph being an automated taxonomy of online information resources, the concept graph comprising information on a plurality of concepts of the online information resources and further comprising relationships among the plurality of concepts; generating a set of concepts, the set of concepts comprising the plurality of concepts matching the search query, the plurality of concepts associated with the search results and the related concepts; clustering the set of concepts into a plurality of grouping of top concepts, wherein the relationships among the plurality of concepts in the concept graph define a boundary of the grouping of top concepts, wherein each grouping represents a potential user intention of the user; identifying a potential user intention of the user corresponding to each grouping of top concepts, wherein each potential user intention comprises one aspect of the topic that the user could be looking for, when the user enters the search query; matching each potential user intentions of the user to an advertisement; and delivering each advertisement and the search results to the user. - View Dependent Claims (13, 14, 15)
-
Specification