Method and system for determining confidence in answer for search

Patent metadata

Field	Value
Publication number	US-9239879-B2
Application number	US-49309709-A
Country	US
Kind code	B2
Filing date	Jun 26, 2009
Priority date	Jun 26, 2009
Publication date	Jan 19, 2016
Grant date	Jan 19, 2016

Abstract

In a method for a direct answer for search, a search query is received over a network, one or more answer candidate snippets for the search query are received, with an answer candidate snippet having at least a portion of content available over the network for an answer candidate, one or more answer entities are determined within a selected answer candidate snippet, a frequency of the one or more answer entities found within the one or more answer candidate snippets for the search query is determined, a confidence score is adjusted for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more answer candidate snippets, and at least one answer candidate snippet is sent for a response to the search query.

First claim

What is claimed:

1. A method for determining confidence in an answer in response to search request, comprising:
receiving, at a direct answer computer system, a search query over a network;
receiving, at the direct answer computer system, one or more answer candidate snippets for the search query, wherein an answer candidate snippet comprises at least a portion of content available over the network for an answer candidate;
determining, at the direct answer computer system, one or more answer entities within a selected answer candidate snippet;
determining, at the direct answer computer system, a frequency of the one or more answer entities found within the one or more answer candidate snippets for the search query;
adjusting, at the direct answer computer system, a confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more answer candidate snippets;
querying, at the direct answer computer system, a reference site database with the search query;
determining, at the direct answer computer system, a plurality of reference site entities within a query result received from the reference site database;
determining, at the direct answer computer system, a frequency of the a plurality answer entities found within the plurality of reference site entities; and
adjusting, at the direct answer computer system, the confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more reference site entities, wherein the confidence score is increased each time an answer entity in a selected answer candidate snippet is found within another answer candidate snippet;
ranking, at the direct answer computer system, the candidate snippets basd on the confidence score as adjusted; and
sending, at the direct answer computer system, at least one answer candidate snippet for a response to the search query.

2. The method of claim 1 , wherein the answer candidate snippet comprises a content identifier, a title, and a description.

3. The method of claim 1 , further comprising:
determining, at the direct answer computer system, one or more search term entities within one or more related search queries;
determining, at the direct answer computer system, a frequency of the one or more answer entities found within the one or more search term entities; and
adjusting, at the direct answer computer system, the confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more search term entities.

4. A method for determining confidence in an answer in response to search request, comprising:
receiving, at a direct answer computer system, a search query over a network;
receiving, at the direct answer computer system, one or more answer candidate snippets for the search query, wherein an answer candidate snippet comprises at least a portion of content available over the network for an answer candidate;
determining, at the direct answer computer system, one or more answer entities within a selected answer candidate snippet;
sending, from the direct answer computer system, a query request to a behavioral database, the query request comprising a request for a plurality of related search queries to the search query;
receiving, at the direct answer computer system, a query result from the behavioral database, the query result comprising the plurality of related search queries to the search query;
determining, at the direct answer computer system, a plurality of search term entities within the one or more related search queries;
determining, at the direct answer computer system, a frequency of the plurality of answer entities found within the one or more search term entities;
adjusting, at the direct answer computer system, a confidence score for the selected answer candidate in accordance with the frequency of the one or more entities found within the one or more search term entities;
ranking the candidate snippets based on the confidence score as adjusted; and
sending at least one answer candidate snippet for a response to the search query.

5. The method of claim 4 , wherein the answer candidate snippet comprises a content identifier, a title, and a description.

6. The method of claim 4 , further comprising:
querying a reference site database with the search query;
determining one or more reference site entities within a query result received from the reference site database;
determining a frequency of the one or more answer entities found within the one or more reference site entities; and
adjusting the confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more reference site entities.

7. The method of claim 4 , wherein the confidence score for the selected answer candidate is increased each time at least one of the one or more entities is found within the one or more search term entities.

8. The method of claim 4 , wherein the confidence score is increased each time an answer entity in a selected answer candidate snippet is found within another answer candidate snippet.

9. A system for determining confidence in an answer in response to search request, comprising:
an interface to receive a search query over a network and send at least one answer candidate snippet for a response to the search query;
an information retrieval module to receive one or more answer candidate snippets for the search query, wherein an answer candidate snippet comprises at least a portion of content available over the network for an answer candidate;
a confidence module to determine one or more answer entities within a selected answer candidate snippet and a frequency of the one or more answer entities found within the one or more answer candidate snippets for the search query, and adjust a confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found with the one or more answer candidate snippets;
a reference site database to query with the search query, wherein the confidence module determines one or more reference site entities within a query result received from the reference site database, a frequency of the one or more answer entities found within the one or more reference site entities, and adjusts the confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more reference site entities, wherein the confidence score is increased each time an answer entity in a selected answer candidate snippet is found within another answer candidate snippet; and
a presentation module ranking the candidate snippets based on the confidence score as adjusted.

10. The system of claim 9 , wherein the answer candidate snippet comprises a content identifier, a title, and a description.

11. The system of claim 9 , wherein the confidence module determines one or more search term entities within one or more related search queries, a frequency of the one or more answer entities found within the one or more search term entities, and adjusts the confidence score for the selected answer candidate in accordance with the frequency of the one or more answer entities found within the one or more search term entities.

12. A system for determining confidence in an answer in response to search request, comprising:
an interface to receive a search query over a network and send at least one answer candidate snippet for a response to the search query;
an inf

Assignees

• Pan Yufei
• Li Hui
• Sarma Justin
• Soukal David
• Signorini Alessio
• Gerasoulis Apostolos
• Imielinski Tomasz
• Iac Search & Media Inc

Inventors

• Pan Yufei
• Li Hui
• Sarma Justin
• Soukal David
• Signorini Alessio
• Gerasoulis Apostolos
• Imielinski Tomasz

Classifications

• G06F17/3053

  Physics · mapped topic

• G06F17/30864 Primary

  Physics · mapped topic

• G06F16/951 Primary

  Indexing; Web crawling techniques · CPC title

• G06F16/953 Primary

  Querying, e.g. by the use of web search engines · CPC title

• G06F16/24578

  using ranking · CPC title