On Using Semantic Web Query Languages for Semantic Web Services Provisioning José María García, Carlos Rivero, David Ruiz and Antonio Ruiz-Cortés Dept. Lenguajes y Sistemas Informáticos, University of Seville ETS Ingeniería Informática, Av. Reina Mercedes s/n, 41012 Sevilla, Spain Abstract— Although there are several approaches to dis- cover Semantic Web Services based on Description Logics reasoning, the use of standard Semantic Web query lan- guages for this task is not so widely spread, partly because service discovery involves some issues that these languages do not usually deal with, such as complex matching, results ranking or interoperability. In this work we analyze the suit- ability of existing query languages to perform provisioning tasks (namely discovery, ranking and selection) within a Se- mantic Web Services scenario. Additionally, the requirements a Semantic Web query language has to fulfill in order to be used within a provisioning scenario are enumerated, giving some insights into how to extend current query languages to do so. Furthermore, an analysis of current provisioning proposals achievement of those requirements is presented. Keywords: Service Discovery, Semantic Web Services, Ontology Languages, Query Languages, Semantic Web. 1. Introduction Once a service has been published and made available from a repository, potential users can fetch for desired services. This fetching, referenced as service provisioning or procurement [1], involves different sequential tasks, namely discovery, ranking and selection. Firstly, services that fulfill the user requirements are discovered. Secondly, those ser- vices are ranked with respect to user preferences. Finally, the best ranked service is selected so it can be executed later on. Usually, semantic discovery is considered as a functional filter, because at this stage the user is looking for a service that provides a requested functionality. Current discovery approaches present matchmaking algorithms that are highly coupled with the service representation formalism, often based on Descriptions Logics [2]–[7]. These proposals de- fines matching degrees that measure the similarity between the user requirements and the available service descriptions. Furthermore, current proposals rely on service descriptions and user preferences defined using OWL- S [8] and WSMO [9] ontologies. Ranking and selection often involves non-functional prop- erties defined over services, e.g. cost or availability. These properties are used to obtain a ranking of discovered ser- vices, so the best service, in terms of user preferences which are based on said properties, can be selected. Current propos- als provide ontologies to express non-functional properties about services that are used within ad-hoc ranking algo- rithms [10]–[13]. As with discovery approaches, ranking and selection proposals have a high coupling between preference description formalisms and algorithms used to perform these tasks. The question that arise in this provisioning scenario is: why current proposals are not using a Semantic Web query language to perform discovery, ranking and selection? One reason can be found at the level of maturity of these query languages so that, until recently there has not been a standard query language for the Semantic Web. Nevertheless, the sometimes complex reasoning needed to match services and user preferences conforms a key feature that current query languages do not completely support, especially RDF-based languages. In this paper we depict the requirements a query language has to fulfill in order to be used to discover and rank Semantic Web Services (SWS), thoroughly analyzing the suitability of current query languages for provisioning tasks, and discussing extensions that make these languages compliant with the enumerated requirements, at least in part. The rest of the paper is structured as follows. In Sec. 2 ex- isting query languages for the Semantic Web are described. Then, in Sec. 3 an analysis of the requirements for these query languages to support SWS provisioning is presented, along with a discussion about current proposals on the topic. Finally, in Sec. 4 we sum up our contributions, and discuss our conclusions. 2. Query Languages for the Semantic Web One of the most important features of the Semantic Web is that it separates the data information from the schema model to be applied to this information [14]. In the Semantic Web field, the W3C recommends RDF (Resource Descrip- tion Framework) [15] as the data model, and RDFS (RDF Schema) [16] or OWL (Web Ontology Language) [17] as the schema model. OWL is divided in three increasingly expressive sub- languages: Lite, DL and Full. DL stands for Description Logics, which is a logical formalism that provides the theoretical foundations for Semantic Web ontologies [18],