Multidimensional Data Mining for Discover Association Rules in Various Granularities

Abstract

Data Mining is one of the most significant tools for discovering association patterns that are useful for many knowledge domains. Yet, there are some drawbacks in existing mining techniques. The three main weaknesses of current datamining techniques are: 1) re-scanning of the entire database must be done whenever new attributes are added because current methods are based on flat-mining using pre-defined schemata. 2) An association rule may be true on a certain granularity but fail on a smaller ones and vise verse. This may result in loss of important association rules. 3) Current methods can only be used to find either frequent rules or infrequent rules, but not both at the same time. This research proposes a novel data schema and an algorithm that solves the above weaknesses while improving on the efficiency and effectiveness of data mining strategies. Crucial mechanisms in each step will be clarified in this paper. This paper also presents a benchmark which is used to compare the level of efficiency and effectiveness of the proposed algorithm against other known methods. Finally, this paper presents experimental results regarding efficiency, scalability, information loss, etc. of the proposed approach to prove its advantages.