Fuzzy logic hybrid model with semantic filtering approach for pseudo relevance feedback-based query expansion

Abstract

Individual query expansion term selection methods have been widely investigated in an attempt to improve their performance. Each expansion term selection method has its own weaknesses and strengths. To overcome the weaknesses and utilize the strengths of individual methods, this paper combined multiple term selection methods. In this paper, initially the possibility of improving the overall performance using individual query expansion (QE) term selection methods are explored. Secondly, some well-known rank aggregation approaches are used for combining multiple QE term selection methods. Thirdly, a new fuzzy logic-based QE approach that considers the relevance score produced by different rank aggregation approaches is proposed. The proposed fuzzy logic approach combines different weights of each term using fuzzy rules to infer the weights of the additional query terms. Finally, Word2vec approach is used to filter semantically irrelevant terms obtained after applying the fuzzy logic approach. The experimental results demonstrate that the proposed approaches achieve significant improvements over each individual term selection method, aggregated method and related state-of-the-art method.