A Hybrid Query Expansion Method for Effective Bengali Information Retrieval

Automatic query expansion has been extensively studied in information retrieval research as a technique for dealing with the word mismatch problem that occurs when the query words do not match with the words in the documents. In this paper, we propose a hybrid query expansion (QE) method which has two phases: (1) candidate expansion term generation phase and (2) expansion term selection phase. In the candidate expansion term generation phase, the statistical, lexical, and semantic methods are combined to automatically generate the candidate expansion terms related to a given query. In the expansion term selection phase, the candidate terms are ranked based on their contextual and semantic relatedness with the given query. To prove the effectiveness of the proposed hybrid query expansion method, we incorporate the proposed method in the Okapi BM25 IR model and then we test the retrieval performance of the enhanced Okapi BM25 IR model on the FIRE dataset which is a benchmark dataset for Bengali information retrieval. The experimental results demonstrate the effectiveness of the proposed hybrid query expansion method. The proposed hybrid query expansion method outperforms the basic Okapi BM25 model by a wide margin.