Towards an Efficient Datalog Based Evaluation of the FSAQL Query Language Afef Bahri, Rafik Bouaziz, and Fa¨ ıez Gargouri University of Sfax, MIRACL Laboratory, Route de M’harza Km 1.5; B.P.: 1030 Sfax: 3018, Tunisia afef.bahri@gmail.com, {raf.bouaziz,faiez.gargouri}@fsegs.rnu.tn Abstract. The resources in the Semantic Web are described using par- ticular metadata called “Semantic annotations”. A semantic annotation is a particular case of annotation which refers to ontology. The Web content is, for the most part, subject to uncertainty or imperfection. FSAQL is a fuzzy query language proposed to query semantic annota- tions defined with fuzzy RDFS. The efficiency of Datalog systems to query large amount of data has been proven in the literature. We pro- pose in this paper an efficient Datalog based approach for the evaluation of the FSAQL query language. The particularity of our approach con- sists on the fact that we use crisp Datalog programs instead of fuzzy ones. In fact, there is no known implementation of fuzzy Datalog sys- tems and the use of crisp Datalog allows the interporability of our query language. Two approaches have been proposed for a correct mapping of fuzzy RDFS to crisp Datalog programs. The defuzzification approach defines crisp “α-cut” classes and properties and maps them to crisp Datalog predicates. The skolemisation approach represents fuzzy classes and properties with crisp Datalog predicates having the same names. The membership degrees are then defined as terms of theses predicates. The two approaches are implemented and evaluated using the F lora-2 Datalog system. 1 Introduction The Semantic Web is an infrastructure that enables the interchange, the in- tegration and the reasoning about information on the Web. One of the most important layer of the Semantic Web is hence the ontology layer. In fact, it pro- vides the conceptual structure that can be used to describe web resources and opens up opportunities for automated information processing. In the Semantic Web, the resources are described using particular metadata called “Semantic annotations”. Their purpose is to assign to objects of a resource a meaning us- ing ontology terms. Many formal standard representations like RDF, RDFS and OWL may be used to allow the representation of web resources in a common and unified way. All these formalisms are based on crisp logic and suppose an exact definition of the resources. In the real world, however, information is subject to imperfection. Also, a web resource is itself source of imperfection and by this way, Z. Lacroix, E. Ruckhaus, and M.-E. Vidal (Eds.): RED 2012, LNCS 8194, pp. 150–180, 2013. c  Springer-Verlag Berlin Heidelberg 2013