Semantics-Based Classification of Rule Interestingness Measures

Abstract

Assessing rules with interestingness measures is the cornerstone of successful applications of association rule discovery. However, as numerous measures may be found in the literature, choosing the measures to be applied for a given application is a difficult task. In this chapter, the authors present a novel and useful classification of interestingness measures according to three criteria: the subject, the scope, and the nature of the measure. These criteria seem essential to grasp the meaning of the measures, and therefore to help the user to choose the ones (s)he wants to apply. Moreover, the classification allows one to compare the rules to closely related concepts such as similarities, implications, and equivalences. Finally, the classification shows that some interesting combinations of the criteria are not satisfied by any index.

Characterization of Interestingness Measures Using Correlation Analysis and Association Rule Mining

“…Exponential development of semantic web, numerous linked data from social community, companies or end-users trigger new potential to connect and join data in order to share and reveal hidden relationship or "semantic association" [22]. Blanchard J. et al [18] studied in semantics-based classification of rule interestingness measures. The experiments showed that according to three criteria such as the subject, the scope, and the nature of the measure are novel and useful for classification of interestingness measures.…”

“…Most existing methods [12], [13], [15]- [18] used real data for investigation and their techniques are clustering and correlation analysis with/without support-based pruning or table standardization. As a different approach, instead of real datasets, one of our previous works [19] investigated interestingness measures using a synthesis dataset but it is not systematic with a relatively small dataset.…”

“…Generally selecting a suitable measure [12]- [14] requires domain and knowledge expertise in order to identify targeted measures or modify specific constraints. Unlike most existing methods [12], [13], [15]- [18] that used real data for investigating, our previous work [19] performed an investigation to characterize interestingness measures using a synthesis dataset. However, the investigation is not systematic and only a small number of cases are used for comparison.…”