Linguistic Decision Tree Induction
Qin Zengchang, Jonathan Lawry · Bristol Research (University of Bristol) · 2003
Label Semantics is a random set based framework for modeling imprecise concepts where the degree of appropriateness of a linguistic expression as a description of a certain value is measured in terms of how the set of appropriate labels for that value varies across a population. An approach to decision tree induction based on this framework was studied. A new decision tree learning algorithm was proposed and its performance applied in real-world data sets was compared with the C4.5 algorithm.