Association rule mining based on concept lattice

Abstract

From the view of concept formation, the paper researched the theories and methods of data mining based on concept lattice theory. The process of knowledge discovery from database may be understood as the process of concept formation from database. The concept lattice theory provides such a formal tool to reflect the process of concept formation. Through this theory, the intension and extension can be formal expressed, the analysis objects can be converted to formal context, and from these formal contexts, the concepts in different hierarchies and their relations can be extracted, and the aim of data mining can be achieved. The algorithm of association rule mining includes two steps: the construction of concept lattice and the production of association rule. The paper produced a fast construction algorithm of incremental concept lattice based on indexed tree. The actual experiment results proved: the algorithm of this paper is faster and more efficient than the traditional association rule mining algorithms-Apriori algorithms, and the algorithm can automated delete the redundant rules, can carry the aim of association rule automated simplified.