α-partitioning algorithm: Vertical partitioning based on the fuzzy graph

Abstract

The issue of vertical partitioning was extensively studied with many proposed approaches, which can be applied to the areas where the match between data and transactions affects performance. Vertical partitioning is the process of producing groups of attributes, called fragments, which are composed of attributes with high affinity each other. Though a vertical partitioning method should be able to generate arbitrary n fragments as well as all meaningful fragments, previous methods have some limitations to support both of them. In this paper we propose an efficient and flexible vertical partitioning method based on a fuzzy graph. The method can not only generate all meaningful fragments at a time but also support n-ary partitioning without any complex mathematical computations.