Combining RDF and SPARQL with CP−Theories to Reason about Preferences in a Linked Data Setting

Abstract

Preference representation and reasoning play a central role in supporting users with complex and multi-factorial decision processes. In fact, user tastes can be used to filter information and data in a personalized way, thus maximizing their expected utility. Over the years, many frameworks and languages have been proposed to deal with user preferences. Among them, one of the most prominent formalism to represent and reason with (qualitative) conditional preferences (CPs) are conditional preference theories (CP-theories). In this paper, we show how to combine them with Semantic Web technologies in order to encode in a standard SPARQL 1.1 query the semantics of a set of CP statements representing user preferences by means of RDF triples that refer to a “preference” OWL ontology. In particular, here we focus on context-uniform conditional (cuc) acyclic CP-theories [Artif. Intell. 175 2011, 1053–1091]. The framework that we propose allows a standard SPARQL client to query Linked Data datasets, and to order the results of such queries relative to a set of user preferences.