Weighted horn clause: Extending SWRL to model antecedents’ importance and handle missing data

Ontologies are formal and explicit specifications of shared conceptualisations. Horn clause (HC) rules may enrich an ontology for modelling complex knowledge and enhancing its expressiveness. While the Semantic Web Rule Language (SWRL) offers a human-readable syntax for incorporating HC into an ontology, it is often too rigid for specific applications, lacking the reasoning nuances needed by domain experts. Additionally, SWRL struggles to handle missing data when inferring new axioms. To address these issues, we propose Weighted Horn Clauses (WHC), an extension of SWRL that incorporates weights to model the importance of antecedent atoms, allowing for more flexible reasoning. WHC syntax and model-theoretical semantics are detailed. We also show how WHC handles missing data when inferring new knowledge in backward and forward chaining strategies. Finally, we propose an open-source prototype reasoner for WHC rules, which is evaluated against SWRL through qualitative and quantitative evaluations. These evaluations illustrate the relevance and feasibility of WHC.