A relatedness-based data-driven approach to determination of interestingness of association rules
Rajesh Natarajan, B. H. Shekar · 2005
The presence of unrelated or weakly related item-pairs can help in identifying Interesting Association Rules (ARs) in a market basket. We introduce three measures for capturing the extent of mutual interaction, substitutive and complementary relationships between two items. Item-relatedness, a composite of these relationships, can help to rank interestingness of an AR. The approach presented, is intuitive and can complement and enhance classical objective measures of interestingness.