Bc: a Rst-order Bayesian Classiier Content Areas: Machine Learning

In this paper we present 1BC, a rst-order Bayesian Classiier. While the propositional Bayesian Classiier makes the naive Bayes assumption of statistical independence of atomic features (one attribute taking on a particular value) given the class value, it is not immediate which atomic features to use in the rst-order case, where features may be constructed from arbitrary numbers of literals. Our approach is to view individuals as structured terms, and to distinguish between structural predicates referring to subterms (e.g. atoms from molecules), and properties applying to one or several of these subterms (e.g. a bond between two atoms). An atomic rst-order feature then consists of zero or more structural predicates and one property. 1BC has been implemented in the context of the rst-order descriptive learner Tertius, and we describe several experiments demonstrating the viability of our approach.