Method and system for facilitating the refinement of data queries

• US 6,678,679 B1
• Filed: 10/10/2000
• Issued: 01/13/2004
• Est. Priority Date: 10/10/2000
• Status: Expired due to Term

• Current Assignee: Relativity ODA LLC
• Original Assignee: Science Applications International Corporation
• Inventors: Bradford, Roger Burrowes
• Primary Examiner(s): Alam, Shahid Al
• Application Number:
  US09/684,330
• Time in Patent Office: 1,190 Days
• Field of Search: 707/1, 707/2, 707/3, 707/5, 707/6, 707/100, 707/501.1, 707/4, 704/9, 715/501.1
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3326 using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

Methods and systems for forming and testing hypotheses regarding improving the correspondence between information needs and data retrieved in response to those needs. In a preferred embodiment, user feedback regarding relevancy of retrieved data is used in combination with the technique of latent semantic indexing to independently facilitate both hypothesis formation and testing.

Claims

1. In an information retrieval system having a user, a current query, a search engine that is responsive to the current query to retrieve data generally relevant to the current query from a data collection, and data generally relevant to the current query retrieved from a target data collection, a method for refining the current query, the method comprising:

indexing a subset of the retrieved data into a vector space;

receiving information regarding the relevancy of a subset of the retrieved data;

ranking a subset of the retrieved data in accordance with the relevancy information;

forming at least one hypothesis regarding modifications to the current query, each hypothesis based at least in part on one or more of;

the rankings and

analysis of locations of the data in the vector space;

creating at least one candidate query in accordance with at least one of the formed at least one hypothesis;

applying at least one of the at least one candidate queries to the retrieved data, employing a search engine responsive to the at least one applied candidate query to retrieve data relevant to the at least one applied candidate query;

ranking the data retrieved in response to the applied candidate query in accordance with the received relevancy information;

comparing the ranking of data retrieved in response to the applied candidate query and the ranking of data retrieved in response to the current query with the received relevancy information

choosing the query which produces results more closely matching the ranking of the retrieved data as a refined query

thereby refining the current query.

2. A method for testing a hypothesis for refining a current query, the method comprising:

indexing a set of data into a vector space

receiving information regarding the relevancy, with respect to information needs, of a subset of the data;

ranking a subset of the data received in response to a current query, based on the degree of responsiveness to the relevancy information;

modifying the current query to incorporate modifications derived from a hypothesis for refining the current query;

applying the modified query to a subset of the data;

ranking a subset of the data retrieved in response to the modified query based at least in part on;

location of the data in the vector space, and

a subset of the received relevancy information;

determining

which of;

the ranking of the subset of data retrieved in response to the current query and

the ranking of the subset of data retrieved in response to the modified query

better corresponds to the relevancy information

thereby testing the hypothesis that incorporation of the modifications improves the correspondence between the query and the received relevancy information.

3. A method for testing a hypothesis for refining a strategy for ranking the results of a data retrieval operation, the method comprising:

indexing a set of data into a vector space;

receiving information regarding the relevancy, with respect to information needs, of a subset of the data;

ranking a subset of the data based on a current ranking strategy;

modifying the ranking strategy to incorporate at least one modification based at least in part on;

location of the data in the vector space, and

a subset of the relevancy information;

ranking a subset of the retrieved data based on the modified ranking strategy determining which of;

the ranking of the subset of retrieved data based on the current ranking strategy and

the ranking of the subset of retrieved data based on the modified ranking strategy

better corresponds to the relevancy information

thereby testing the hypothesis that incorporation of the at least one modification into the current ranking strategy improves the correspondence between the ranking and the received relevancy information.