Linguistic FOIL and Multiple Attribute Hierarchy for Decision Making

Rule induction has been well accepted as a transparent learning system for its interpretable rules in decision making. In this chapter, a new linguistic rule induction algorithm is proposed by incorporating label semantics into the FOIL (First-Order Inductive Learning) algorithm. The latter is a well-known first-order rule induction algorithm in Inductive Logical Programming (ILP) proposed by Quinlan^[1]. The new algorithm is guided by information based heuristics in accordance with label semantics. Some benchmark problems are tested and the results show that the new algorithm has comparable accuracy with linguistic rules. In the second part of this chapter, the multiple attribute decision making based on linguistic decision trees is given and illustrated by using an example.