Accepted as Short Paper at CIKM ’18: ACM International Conference on Information and Knowledge Management[1].

Abstract. Query Expansion (QE) is a well-known technique aimed at enhancing recall by expanding user queries with additional terms (e.g., synonyms, plurals, acronyms, etc.). State-of-the art solutions select the most suitable expansions by employing supervised learning techniques. Most of them, however, neglect to consider the processing cost of the expanded queries and may generate effective expansions which are very expensive to process. The goal of this paper is to propose a query expansion framework based on structured expanded queries and efficiency-aware term selection strategies, with the purpose of enabling efficient QE in scenarios with tight time constraints. In particular the proposed expansion selection strategies aim at capturing the efficiency and effectiveness of the expansions candidates, as well as the multi-dimensional dependencies among them. We evaluate our proposal by conducting a comprehensive experimental assessment on datasets from a real-world search engine and public TREC data. Results confirm that our approach compared to the state of the art leads to a remarkable improvement in efficiency, i.e., up to one order of magnitude reduction of retrieval time, with only a small loss in terms of effectiveness.