ISA-An Incremental Hyper-Sphere Approach for Efficiently Solving Complex Vague Queries

This paper presents an enhanced extension of the Vague Query System (VQS) [14, 15] to solve complex vague queries more efficiently and more generally. The VQS is an extension to conventional database systems and can work on top of them to return records semantically close to user's query. It represents semantic information of arbitrary attributes by mapping them to Euclidean space and this information is invisible to the users. Answering a complex vague query requires parallel searching on multiple feature spaces of attributes and these spaces are usually multidimensional. The proposed approach in [15] for solving complex vague queries is not general and has weaknesses lead to degenerate performance of the system. The new approach introduced in this paper addresses all those defects and significantly improves the search performance in both concepts of CPU-time and IO-cost. Our experiments on uniformly distributed data sets, which are managed by the SH-trees [6], have proven the advantages of this new approach, which really makes the VQS full-fledged.