Rule Acquisition in Generalized One-Sided Decision Systems

Zhiyong Hu, Mingwen Shao, Meishe Liang

Abstract: As a major direction of data mining, rule acquisition has been applied and researched widely. However, there are few studies on association rules determined by ordered attributes. In this paper, we propose methods of mining decision rules based on situations where only condition attributes are ordered (generalized one-sided formal decision context) and both the condition and decision attributes are ordered (generalized one-sided ordered formal decision context), to discuss the influence of ordered attributes on decision rules. Moreover, the discussion is carried out to explain relations between decision rules based on generalized one-sided formal decision contexts and decision rules based on generalized one-sided ordered formal decision contexts. Furthermore, the rules based on ordered decision attributes pay more attention to global information while the rules constructed by disordered decision attributes pay more attention to local information through the experimental results and theoretical derivation.