Association discovery
Geoffrey I. Webb · 2010
Association discovery is one of the most studied tasks in the field of data mining. However, far more attention has been paid to how to discover associations than to what associations should be discovered. In this talk Geoff will provide a highly subjective tour of the field. He will highlight shortcomings of the dominant frequent pattern paradigm and illustrate benefits of the alternative top-k approach. He will argue for the value of statistical filtering of associations and discuss some null-hypotheses of widespread application. He will compare the merits of randomization, holdout and within-search approaches to statistical filtering. He will also argue that in many applications it is preferable to find interesting itemsets rather than interesting rules.