Exploiting Embedded Synopsis for Exact and Approximate Query Processing

Abstract

AbstractOrganizing data in structured forms is a mainstream approach to improve the efficiency of query processing on database. For example, a B+Tree index is widely employed to limit the search space for selective queries, speeding up the query processing. In this direction, this paper proposes synopsis embedment, an idea of incorporating synopsis (i.e. precomputed summary or abstraction information) into an existing data structure. Using the embedded synopsis, the query processing may obtain an exact or approximate answer without accessing all the concerned records in the data structure. Thus, the use of the synopsis may reduce the access cost, improving the efficiency of the query processing. This paper presents a design of efficient search algorithms on the synopsis-embedded B+Tree structure and experimentally clarifies its performance superiority.KeywordsSynopsis embedmentData structureApproximate query processing