Mining Non-redundant Information-Theoretic Dependencies between Itemsets

Abstract. We present an information-theoretic framework for mining dependencies between itemsets in binary data. The problem of closure-based redundancy in this context is theoretically investigated, and we present both lossless and lossy pruning techniques. An efficient and scalable algorithm is proposed, which exploits the inclusion-exclusion principle for fast entropy computation. This algorithm is empirically evaluated through experiments on synthetic and real-world data.

• Type: Journal
• Year: 2010
• Where: DAWAK
• Authors: Michael Mampaey
• Keywords: Binary Data, Closure-based Redundancy, DAWAK 2010, Information Technology, Information-theoretic Framework