Evaluation of N-grams Conflation Approach in Text-Based Information Retrieval

Abstract. This paper examines a conflation method based on the N-grams approach and evaluates its performance relative to the results achieved by other techniques such as Porter algorithm and successor variety stemming. In addition to that, an alternative way of enhancing the N-grams method, derived from the concept of inverse frequency weighing, is introduced and evaluated. The experimental results generated using standard collections ADI, CISI and Medlars show an improvement over the traditional conflation methods, as well as demonstrate the viability of the introduced inverse frequency multiplier technique.

• Type: Conference
• Year: 2001
• Where: SPIRE
• Authors: S. Kosinov
• Keywords: Conflation Methods, Information Management, Inverse Frequency, Inverse Frequency Weighing, SPIRE 2001