Using Taxonomic Background Knowledge in Propositionalization and Rule Learning

Knowledge representations using semantic web technologies often provide information which translates to explicit term and predicate taxonomies in relational learning. Here we show how to speed up the process of propositionalization of relational data by orders of magnitude, by exploiting such ontologies through a novel refinement operator used in the construction of conjunctive relational features. Moreover, we accelerate the subsequent search conducted by a propositional learning algorithm by providing it with information on feature generality taxonomy, determined from the initial term and predicate taxonomies but also accounting for traditional θ-subsumption between features. This information enables the propositional rule learner to prevent the exploration of useless conjunctions containing a feature together with any of its subsumees and to specialize a rule by replacing a feature by its subsumee. We investigate our approach with a propositionalization algorithm, a deterministic top-down propositional rule learner, and a recently proposed propositional rule learner based on stochastic local search. Experimental results on genomic and engineering data [2] indicate striking runtime improvements of the propositionalization process and the subsequent propositional learning.