Abstract
This chapter presents the different theoretical frameworks, condensed representations, interestingness measures, and biological applications of association rule mining. It presents concepts and frameworks for association rule mining and their applications in biology, especially in genomics and proteomics. The different theoretical frameworks proposed for itemset representation and frequent-itemset extraction are described here. The chapter describes the proposed solutions for reducing sets of extracted association rules to the most relevant and useful rules. This is an important topic in association rule mining as several thousands, and sometimes millions, of association rules can be generated for large databases, with most often numerous redundant information. The different condensed representations of association rules, such as minimal covers, bases, and inference systems, along with their properties are presented in the chapter. It presents subjective and objective interestingness measures that can be used for selecting the most relevant association rules.
