Enhancement of passage scorers by proximity-based term occurrence weighting. (1st January 2013)

• Record Type: Journal Article
• Authors: Liu, Rey-Long; Hu, Rey-Hsing
• Is Part Of: International journal of intelligent information and database systems. Volume 7:Number 6(2013)
• Page Start: 496
• Page End: 515
• Publication Date: 2013-01-01
• Subjects: passage retrieval -- passage scoring -- term proximity -- term occurrence weighting; Database management -- Computer programs -- Periodicals; Information retrieval -- Computer programs -- Periodicals; Information storage and retrieval systems -- Computer programs -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Intelligent agents (Computer software) -- Periodicals; 006.33
• Languages: English
• ISSNs: 1751-5858

Abstract

Given a query and a document, passage retrievers aim at selecting from the document those passages that are relevant to the query. Passage scoring is thus essential for passage retrievers. In this paper, we present a technique named proximity-based term occurrence weighting (PTOW), which employs term proximity information to enhance various kinds of passage scorers. For each occurrence of a query term t at position k in a passage, PTOW increments the weight of the occurrence based on how other query terms appear around k . The occurrence weight is then passed to the passage scorers. Empirical evaluation shows that PTOW significantly enhances two passage scorers. Moreover, when compared with two state-of-the-art techniques that enhance scorers by term proximity information, PTOW performs significantly better as well. The contributions are of practical significance, since many passage scorers have been developed and PTOW may further enhance their performance.