A Bayesian Approach for Semantic Search Based on DAG-Shaped Ontologies

Semantic search has a great potentiality in helping users to make choices, since it appears to outperform traditional keyword-based approaches. This paper presents an ontology-based semantic search method, referred to as influential SemSim (i-SemSim), which relies on the Bayesian probabilistic approach for weighting the reference ontology. The Bayesian approach seems promising when the reference ontology is organized according to a Directed Acyclic Graph (DAG). In particular, in the proposed method the similarity among a user request and semantically annotated resources is evaluated. The user request, as well as each annotated resource, is represented by a set of concepts of the reference ontology. The experimental results of this paper show that the adoption of the Bayesian method for weighting DAG-based reference ontologies allows i-SemSim to outperform the most representative methods selected in the literature.