Method and system of weighted context feedback for result improvement in information retrieval

• US 7,003,513 B2
• Filed: 05/14/2001
• Issued: 02/21/2006
• Est. Priority Date: 07/04/2000
• Status: Expired due to Term

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Geiselhart, Reinhold
• Primary Examiner(s): Ali, Mohammad
• Assistant Examiner(s): Lu, Kuen S.
• Application Number:
  US09/855,863
• Publication Number:
  US 20020069190A1
• Time in Patent Office: 1,744 Days
• Field of Search: 707 1- 5, 707/6, 704 1- 10
• 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: Y10S 707/99934 Query formulation, input preparation, or translation
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A method and a system for re-ranking an existing result set of documents. A user starts a search by entering search term(s). The search term(s) is (are) transferred to a search engine which generates a result set ranked by the search term(s). The search engine, in parallel, automatically retrieves context information from returned result set which is related to the original set of documents. The search engine presents the context information to the user and asks for a feedback. The user performs a weighting of the presented context information in a range from “important” to “non-important”. The result set is then re-ranked with the user-weighted context information to increase the “rank distance” of important and non important documents. The documents that are on top of the list (highest context-weighted ranking value) represent the desired information.

Claims

1. A computer implemented method for re-ranking a set of ranked documents, each ranked document having an absolute ranking value determined based on one or more search terms, the method comprising:

retrieving a plurality of context information based on said one or more search terms from each of said set of ranked documents;

presenting said plurality of context information together with a set of associated ranking criteria to a user, said set of associated ranking criteria being based on discrete ranking levels;

receiving user preferences regarding each of said ranking criteria associated with each of said plurality of context information; and

re-ranking said set of ranked documents based on a new ranking value calculated for each of said ranked documents utilizing a context ranking formula based on said absolute ranking value and said user preferences regarding each of said associated ranking criteria.

2. The computer implemented method of claim 1, wherein retrieving a plurality of context information comprises one or more of extracting lexical affinities, extracting features, and extracting word frequency statistics from said set of ranked documents.

3. The computer implemented method of claim 1, wherein user preferences of said associated ranking criteria are based on a weighting function.

4. The computer implemented method of claim 1, wherein said context ranking formula utilizes the following ranking and weighted ranking equations:

ranking equation

fd(x1, . . . , xn)=Rd if x1, . . . , xn are elements of Td, and

fd(x1, . . . , xn)=0 if x1, . . . , xn are not elements of Td,

wherein Rd is the absolute ranking value of a document “

d”

that results from a search, and Td=(x1, . . . , xn) is a tuple of context terms that are contained in the document “

d”

;

weighted ranking equation

[2a f(x1, . . . , xa) (a b) f(x1, . . ., xa b) (a b c) f(x1, . . . , xa b c)]/(4a 2b c)

wherein the weighted ranking equation calculates the relevance of a document with respect to the context terms x1, . . . , xm when a, b and c are the number of terms that have been assigned a high (a), medium (b) and low (c) relevance and f(x1, . . . , xa), f(x1, . . . , xa b) and f(x1, . . . , xa b c) are partial relevance functions of the document with respect to a subset of the context terms.

5. A computer program product comprising a computer useable medium having a computer readable program for re-ranking a set of ranked documents, each ranked document having an absolute ranking value determined based on one or more search terms, said computer readable program when executed on a computer causes the computer to:

retrieve a plurality of context information based on said one or more search terms from each of said set of ranked documents;

present said plurality of context information together with a set of associated ranking criteria to a user, said set of associated ranking criteria being based on discrete ranking levels;

receive user preferences regarding each of said ranking criteria associated with each of said plurality of context information; and

re-rank said set of ranked documents based on a new ranking value calculated for each of said ranked documents utilizing a context ranking formula based on said absolute ranking value and said user preferences regarding each of said associated ranking criteria.

6. The computer program product of claim 5, wherein an operation to retrieve a plurality of context information comprises one or more of extracting lexical affinities, extracting features, and extracting word frequency statistics from said set of ranked documents.

7. The computer program product of claim 5, wherein user preferences of said associated ranking criteria are based on a weighting function.

8. The computer program product of claim 5, wherein said context ranking formula utilizes the following ranking and weighted ranking equations:

ranking equation

fd(x1, . . . , xn)=Rd if x1, . . . , xn are elements of Td, and

fd(x1, . . . , xn)=0 if x1, . . . , xn are not elements of Td,

wherein Rd is the absolute ranking value of a document “

d”

that results from a search, and Td=(x1, . . . , xn) is a tuple of context terms that are contained in the document “

d”

;

weighted ranking equation

[2a f(x1, . . . , xa) (a b) f(x1, . . . , xa b) (a b c) f(x1, . . . , xa b c)]/(4a 2b c)

wherein the weighted ranking equation calculates the relevance of a document with respect to the context terms x1, . . . , xm when a, b and c are the number of terms that have been assigned a high (a), medium (b) and low (c) relevance and f(x1, . . . , xa), f(x1, . . . , xa b) and f(x1, . . . , xa b c) are partial relevance functions of the document with respect to a subset of the context terms.

9. An apparatus for re-ranking a set of ranked documents, each ranked document having an absolute ranking value determined based on one or more search terms, the apparatus comprising, a search engine which:

retrieves a plurality of context information based on said one or more search terms from each of said set of ranked documents;

presents said plurality of context information together with a set of associated ranking criteria to a user, said set of associated ranking criteria being based on discrete ranking levels;

receives user preferences regarding each of said ranking criteria associated with each of said plurality of context information; and

re-ranks said set of ranked documents based on a new ranking value calculated for each of said ranked documents utilizing a context ranking formula based on said absolute ranking value and said user preferences regarding each of said associated ranking criteria.

10. The apparatus of claim 9, wherein the retrieve module is configured to perform operations to retrieve a plurality of context information, the operations comprising one or more of extracting lexical affinities, extracting features, and extracting word frequency statistics from said set of ranked documents.

11. The apparatus of claim 9, wherein user preferences of said associated ranking criteria are based on a weighting function.

12. The apparatus of claim 9, wherein said context ranking formula utilizes the following ranking and weighted ranking equations:

ranking equation

fd(x1, . . . , xn)=Rd if x1, . . . , xn are elements of Td, and

fd(x1, . . . , xn)=0 if x1, . . . , xn are not elements of Td,

wherein Rd is the absolute ranking value of a document “

d”

that results from a search, and Td=(x1, . . . , xn) is a tuple of context terms that are contained in the document “

d”

;

weighted ranking equation

[2a f(x1, . . . , xa) (a b) f(x1, . . . , xa b) (a b c) f(x1, . . . , xa b c)]/(4a 2b c)

wherein the weighted ranking equation calculates the relevance of a document with respect to the context terms x1, . . . , xm when a, b and c are the number of terms that have been assigned a high (a), medium (b) and low (c) relevance and f(x1, . . . , xa), f(x1, . . . , xa b) and f(x1, . . . , xa b c) are partial relevance functions of the document with respect to a subset of the context terms.