Simple algorithms for predicate suggestions using similarity and co-occurrence

Abstract

When creating Semantic Web data, users have to make a critical choice for a vocabulary: only through shared vocabularies can meaning be established. A centralised policy prevents terminology divergence but would restrict users needlessly. As seen in collaborative tagging environments, suggestion mechanisms help terminology convergence without forcing users. We introduce two domain-independent algorithms for recommending predicates (RDF statements) about resources, based on statistical dataset analysis. The first algorithm is based on similarity between resources, the second one is based on co-occurrence of predicates. Experimental evaluation shows very promising results: a high precision with relatively high recall in linear runtime performance.