Recommendation of Text Tags in Social Applications Using Linked Data Andrea Cal` ı 1,2 , Stefano Capuzzi 3 , Mirko Michele Dimartino 1 , and Riccardo Frosini 1 1 Dept. of Computer Science and Inf. Syst., Birkbeck, University of London, UK 2 Oxford-Man Institute of Quantitative Finance, University of Oxford, UK 3 Dipartimento di Ingegneria dell’Informazione, Universit`a di Brescia, Italy Abstract. We present a recommender system that suggests geo-located text tags by using linguistic information extracted from Linked Data sets available on the Web. The recommender system performs tag matching by measuring the semantic similarity of natural language texts. Our ap- proach evaluates similarity using a technique that compares sentences taking into account their grammatical structure. 1 Introduction Presently, large RDF data sets are available, which offer a large amount of machine-processable information for a wide range of applications. In October 2007, the Linked Open Data (LOD) data sets consisted of over two billion RDF triples, interlinked by over two million RDF links. By September 2011 this had grown to 31 billion RDF triples, interlinked by around 504 million RDF links [4]. In spite of its success, the Semantic Web may contain several poor quality data and numerous unsolved data management problems, related both to the concept as well as the implementation [2,6]. Several dictionaries and Thesauri have also been recently published in RDF format, for instance WordNet and Wiktionary. We have designed a social application that allows end-users to retrieve geo- tagged text snippets which are semantically related to natural language queries. To achieve this goal we exploit the above mentioned Linked Data sets. As a first step, we to compare English words and establish a similarity measure using the properties extracted from the Linked Data resources. With such word similarity measure at hand, we use it as a building block to compare whole text snippets. Sentences are compared by considering their syntactic structure, so as to capture their semantics at a somewhat detailed level. In this paper we describe the main features of our first prototype offering seman- tic search of natural language text tags. In our system, the user is able to submit geo-tagged text snippets, describing anything related to the location, by a mobile application; the application stores the tags, sharing them with other users. A user can query the system using a natural language text snippet, in order to retrieve se- mantically related, as well as geographically close, tags submitted by other users. Q.Z. Sheng and J. Kjeldskov (Eds.): ICWE 2013 Workshops, LNCS 8295, pp. 187–191, 2013. c  Springer International Publishing Switzerland 2013