INFORMATION-RETRIEVAL SYSTEMS, METHODS, AND SOFTWARE WITH CONCEPT-BASED SEARCHING AND RANKING

• US 20090198674A1
• Filed: 08/22/2008
• Published: 08/06/2009
• Est. Priority Date: 12/29/2006
• Status: Active Grant

• Current Assignee: Thomson Reuters Enterprise Centre GmbH (The Woodbridge Co. Ltd.)
• Original Assignee: Thomson Reuters Global Resources
• Inventors: Custis, Tonya, Al-Kofahi, Khalid
• Granted Patent: US 8,321,425 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3346 using probabilistic model

Abstract

To improve traditional keyword based search engines, the present inventors devised, among other things, systems, methods, and software that use word co-occurrence probabilities not only to identify documents conceptually related to user queries, but also to score and rank search results. One exemplary system combines inverse-document-frequency searching with concept searching based on word co-occurrence probabilities to facilitate finding of documents that would otherwise go unfound using a given query. The exemplary system also allows ranking of search results based both on both keyword matching and concept presence, promoting more efficient organization and review of search results.

Claims

1. A system comprising:

a set of target documents; and

means for searching and identifying one or more of the set of target documents as result documents based on a user query, the means for searching and identifying including;

means for identifying one or more first documents based on a set of word co-occurrence probabilities, with the set of word co-occurrence probabilities derived from a set of documents the same as or different than the set of target documents.

2. The system of claim 1, wherein the means for searching and identifying one or more of the set of target documents includes means for ranking the result documents based on query term occurrence information and the set of word co-occurrence probabilities.

3. The system of claim 1, wherein the set of target documents includes case law documents.

4. The system of claim 1, wherein the set of target documents includes corporate electronic mail or news documents.

5-8. -8. (canceled)

9. A method for using a query having one or more query terms to identify a set of one or more documents within a database, the method comprising:

determining for each of one or more documents in the database a score based on;

occurrence of one or more of the query terms in the document; and

occurrence or one or more non-query terms that are known to co-occur with one or more of the query terms in a set of documents; and

displaying one or more of the documents within a search result based on its determined score.

10. The method of claim 9, wherein the score is based on a keyword model and a translation language model.

11. The method of claim 9, wherein the score is based on a sum of a first quantity and a second quantity, the first quantity based at least in part on a count of how many times one of the query terms occurs in the document, and the second quantity based at least in part on a count of how many times one of the non-query terms co-occurs with one of the query terms in the set of documents.