Improving pseudo relevance feedback based query expansion using genetic fuzzy approach and semantic similarity notion

• Published in: Journal of information science Vol. 40; no. 4; pp. 523 - 537
• Main Authors: Bhatnagar, Pragati, Pareek, Narendra
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01.08.2014; Sage Publications; Bowker-Saur Ltd
• Subjects: Exact sciences and technology; Feedback; Fuzzy; Fuzzy logic; Fuzzy set theory; General aspects; Genetics; Information and communication sciences; Information retrieval; Information science. Documentation; Query expansion; Query processing; Relevance feedback; Sciences and techniques of general use; Semantics; Similarity; Studies; pseudo relevance feedback; semantic similarity; Genetic fuzzy algorithm; query expansion; information retrieval; Similarity; Use; Feedback regulation; Query expansion; Information retrieval; Fuzzy logic; Relevance; Improvement; Genetic algorithm; Semantics; Genetics
• ISSN: 0165-5515; 1741-6485
• DOI: 10.1177/0165551514533771

Pseudo relevance feedback-based query expansion is a popular automatic query expansion technique. However, a survey of work done in the area shows that it has a mixed chance of success. This paper captures the limitations of pseudo relevance feedback (PRF)-based query expansion and proposes a method of enhancing its performance by hybridizing corpus-based information, with a genetic fuzzy approach and semantic similarity notion. First the paper suggests use of a genetic fuzzy approach to select an optimal combination of query terms from a pool of terms obtained using PRF-based query expansion. The query terms obtained are further ranked on the basis of semantic similarity with original query terms. The experiments were performed on CISI collection, a benchmark dataset for information retrieval. It was found that the results were better in both terms of recall and precision. The main observation is that the hybridization of various techniques of query expansion in an intelligent way allows us to incorporate the good features of all of them. As this is a preliminary attempt in this direction, there is a large scope for enhancing these techniques.