Determination of interesting rules in FCA using information gain

The difficult job in association rules is to identify the frequent item sets immersed into the huge collection of data. The association rules can be discovered using Formal Concept Analysis (FCA). Several contexts often contain large number of rules and hence interesting rules are required to be determined. With this objective, this paper proposes a method for determining interesting rules in FCA involving many-valued contexts based on Shannon's information entropy (IE) theory. For this purpose we define a gain_lift measure on association rules. The proposed method is illustrated by means of an example available from the field of medical diagnosis.