Architecture for generating responses to search engine queries
First Claim
Patent Images
1. A system, comprising:
- one or more processors; and
one or more media storage devices storing a plurality of components, each component of the plurality of components comprising instructions that are executable by the one or more processors, the plurality of components including;
an input component to receive a query;
an abstraction component to;
map the query to an abstraction hierarchy according to a distribution of terms found in potential answer passages for the query by calculating a similarity value between the query and each of a plurality of queries in the abstraction hierarchy; and
group the query with a most similar query in the abstraction hierarchy based on the similarity value; and
an identification component to map data pairs comprising query-response pairs to the abstraction hierarchy to identify informative terms for a set of nodes in the abstraction hierarchy by determining using a Dunning measure a term-by-occurrence contingency data structure for each term found in possible answer passages for each question in each of the query-response pairs.
2 Assignments
0 Petitions
Accused Products
Abstract
Architecture for generating responses to search engine queries. Question/answer pairs are used to provide a predefined question hierarchy to facilitate a search. Informative terms and one or more questions are associated with each question node of the hierarchy. A question posed is mapped into the hierarchy to the most similar question class node, as determined by passages of the question and/or comparing the informative terms of the question and the question/answer node. Weighted informative terms associated with that node are then returned, and can be used by an automatic question answering system to more effectively find an answer to a question.
41 Citations
40 Claims
-
1. A system, comprising:
-
one or more processors; and one or more media storage devices storing a plurality of components, each component of the plurality of components comprising instructions that are executable by the one or more processors, the plurality of components including; an input component to receive a query; an abstraction component to; map the query to an abstraction hierarchy according to a distribution of terms found in potential answer passages for the query by calculating a similarity value between the query and each of a plurality of queries in the abstraction hierarchy; and group the query with a most similar query in the abstraction hierarchy based on the similarity value; and an identification component to map data pairs comprising query-response pairs to the abstraction hierarchy to identify informative terms for a set of nodes in the abstraction hierarchy by determining using a Dunning measure a term-by-occurrence contingency data structure for each term found in possible answer passages for each question in each of the query-response pairs. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. A method comprising:
-
under control of one or more processors configured with instructions that are executable to perform acts comprising; receiving, by the one or more processors, a query; processing, by the one or more processors, the query for informative terms; mapping, by the one or more processors, the query into an abstraction hierarchy in accordance with a distribution of potential answer terms by calculating a similarity value between the query and each query in the abstraction hierarchy to discriminate between a first set of potential answer terms in the abstraction hierarchy that include answers from a second set of potential answer terms in the abstraction hierarchy that do not include the answers; grouping the query with a most similar query in the existing abstraction hierarchy based on the similarity value; and
mapping data pairs comprising query-response pairs to the abstraction hierarchy to identify informative terms for a set of nodes in the abstraction hierarchy by determining, using a Dunning measure, a term-by-occurrence contingency data structure for each term found in possible answer passages for each question in each of the query-response pairs. - View Dependent Claims (17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27)
-
-
28. A method comprising:
-
under control of one or more processors configured with instructions that are executable to perform acts comprising; receiving a question; extracting a term from the question; generating a term-by-occurrence contingency data structure of potential answers from a question/answer hierarchy by calculating a similarity value between the question and questions in the question/answer hierarchy, the term-by-occurrence contingency data structure including the potential answers in which the term occurs; testing the data structure of potential answers for a discriminating term; storing the discriminating term as an informative term using a Dunning measure; ranking a plurality of the informative terms by relevance to a potential answer to the question based on the similarity value; and utilizing one of the informative terms to find the potential answer. - View Dependent Claims (29)
-
-
30. A memory storage device including instructions executable by one or more processors to perform acts comprising:
-
receiving an existing abstraction hierarchy of nodes of questions and corresponding informative terms; receiving a new question; generating informative terms for the new question; calculating a similarity value between the new question and every question in the existing abstraction hierarchy, the similarity value discriminating between a first set of questions in the existing abstraction hierarchy that include similar questions to the new question from a second set of questions in the existing abstraction hierarchy that do not include similar questions to the new question; mapping data pairs comprising question-answer pairs to the existing abstraction hierarchy to identify informative terms for a set of nodes in the abstraction hierarchy by determining, using a Dunning measure, a term-by-occurrence contingency data structure for each term found in possible answer passages for each question in each of the question-answer pairs; and grouping the new question with the most similar question in the existing abstraction hierarchy based on the similarity value. - View Dependent Claims (31, 32)
-
-
33. A system comprising:
-
one or more processors; and one or more media storage devices storing instructions that are executable by the one or more processors to perform acts comprising; receiving a query; processing the query to identify informative terms; mapping the query into an abstraction hierarchy in accordance with a plurality of potential answers for the query; grouping the query with a most similar query in the abstraction hierarchy based on the similarity value; and mapping data pairs comprising query-response pairs to the abstraction hierarchy to identify informative terms for a set of nodes in the abstraction hierarchy by determining, using a Dunning measure, a term-by-occurrence contingency data structure for each term found in possible answer passages for each question in each of the query-response pairs. - View Dependent Claims (34, 35, 36, 37, 38, 39, 40)
-
Specification