Abstract
Query refinement techniques enable database systems to automatically adjust a submitted query so that its result satisfies some specified constraints. While current techniques are fairly successful in generating refined queries based on cardinality constraints, they are rather oblivious to the (dis)similarity between the input query and its corresponding refined version. Meanwhile, enforcing a similarity-aware query refinement is a rather challenging task as it would require an exhaustive examination of the large space of possible query refinements. To address this challenge, we propose a novel scheme for efficient Similarity-aware Query Refinement (SAQR). SAQR aims to balance the tradeoff between satisfying the cardinality and similarity constraints imposed on the refined query so that to maximize its overall benefit to the user. To achieve that goal, SAQR implements efficient strategies to minimize the costs incurred in exploring the available search space. In particular, SAQR utilizes both similarity-based and cardinality-based pruning techniques to bound the search space and quickly find a refined query that meets the user expectations. Our experimental evaluation shows the scalability exhibited by SAQR under various workload settings, and the significant benefits it provides.