Query expansion (QE) is an efficient method for enhancing the efficiency of information retrieval system. In this work, we try to capture the limitations of pseudo-feedback based QE approach and propose a hybrid approach for enhancing the efficiency of feedback based QE by combining corpus-based, contextual based information of query terms, and semantic based knowledge of query terms. First of all, this paper explores the use of different corpus-based lexical co-occurrence approaches to select an optimal combination of query terms from a pool of terms obtained using pseudo-feedback based QE. Next, we explore semantic similarity approach based on word2vec for ranking the QE terms obtained from top pseudo-feedback documents. Further, we combine co-occurrence statistics, contextual window statistics, and semantic similarity based approaches together to select the best expansion terms for query reformulation. The experiments were performed on FIRE ad-hoc and TREC-3 benchmark datasets. The statistics of our proposed experimental results show significant improvement over baseline method.
