Keynote talk #3: Recent methods in mining patterns and association rules

Abstract

Mining patterns and rules is one of the most common topics in data mining. It has been applied to a vast variety of databases such as transactional databases, sequence databases, graph databases, and so on. Additionally, pattern and rule mining has been shown its practical applications in various areas like supermarket, finance, healthcare, education, bioinformatics, and so on. This talk begins with the applications of frequent (closed) itemset lattice to: — mine traditional association rules: build and use a frequent itemset lattice to discover traditional association rules — mine non-redundant association rules: build and use a modified frequent itemset lattice to generate non-redundant association rules — mine most generation association rules: build and use a frequent closed itemset lattice to generate most generation association rules — mine class association rules (CARs) It then presents the dynamic bit vector structure and its applications in mining frequent patterns from transactional databases and frequent (inter)-sequences from sequence databases. Some methods in mining patterns from quantitative databases such as mining high utility itemsets, mining frequent weighted itemsets, etc are also mentioned. Finally, this talk introduces an application of association rule mining to classification such as mining all CARs, mining CARs with constraints, etc.