标题：DIFP: A candidate facet generation optimization method towards keyword-driven analytical processing
作者：Sun, Ming ;Li, Jing ;Wang, Xinjun ;Peng, Zhaohui
作者机构：[Sun, Ming ;Li, Jing ;Wang, Xinjun ;Peng, Zhaohui ] School of Computer Science and Technology, Shandong University, Jinan 250101, China
来源：Journal of Computational Information Systems
关键词：Candidate facet generation; Document-based attribute instance facet processing; Driven analytical processing; Vector space model
摘要：Candidate facet generation, which is responsible for computing the aggregation values for attributes and attribute instances of every dimension and sorting, is an important stage in keyword-driven analytical processing (KDAP). If there are too many attributes in the dimension or the instance table is too large, the efficiency of the candidate facet generation will reduce significantly. This paper proposes a candidate facet generation optimization method based on document-based attribute instance facet processing (DIFP) to improve the generation efficiency. Firstly, we process the candidate subspace to generate partition subspaces. Secondly, the partition subspace is taken as a document set, every instance facet of which is considered as a super document. At last, we compute the similarity between the query and document by vector space model and select the most hopeful instance facets to aggregation computing. The method reduces the aggregation calculation number of candidate instance facets and the experiments prove it can improve the query efficiency of KDAP system significantly with the attribute number or table size increasing. Copyright © 2013 Binary Information Press.