Ontology driven resource discovery in a Bluetooth based m-marketplace Michele Ruta Politecnico di Bari via Re David 200 I-70125 Bari, Italy Email: m.ruta@poliba.it Tommaso Di Noia Politecnico di Bari via Re David 200 I-70125 Bari, Italy Email: t.dinoia@poliba.it Eugenio Di Sciascio Politecnico di Bari via Re David 200 I-70125 Bari, Italy Email: disciascio@poliba.it Giacomo Piscitelli Politecnico di Bari via Re David 200 I-70125 Bari, Italy Email: piscitel@poliba.it Abstract—We present a semantic-based approach to resource retrieval in an m-commerce scenario. We enhance the original Bluetooth Service Discovery Protocol by integrating a “semantic layer” within the application level of the standard. Given a user request, this layer makes possible a matchmaking process exploiting the semantics of the resources descriptions exposed by a hotspot. It computes the match degree between the request and the goods available in the m-marketplace taking into account both their ontology-based descriptions and “classical” attributes such as price difference, availability, quantity. I. INTRODUCTION New mobile architectures allow for stable networked links from almost everywhere, and more and more people make use of information resources for work and business purposes on mobile systems. Although technological improvements in the standardization processes proceed rapidly, many challenges, mostly aimed at the deployment of value added services on mobile platforms, are still unsolved. In particular the evolution of wireless-enabled handheld devices and their capillary dif- fusion have increased the need for more sophisticated Service Discovery Protocols (SDPs). Here we present an approach which enhances Bluetooth SDP, to provide m-commerce resources to the users within a piconet, extending the basic service discovery with semantic capabilities. In particular we exploit and enhance the SDP in order to identify generic resources rather than only services. We have integrated a “semantic layer” within the application level of the standard Bluetooth stack in order to enable a simple interchange of semantically annotated information between a mobile client performing a query and a server exposing available resources. We adopt a simple piconet configuration where a stable networked zone server, equipped with a Bluetooth interface, collects requests from mobile clients and hosts a semantic facilitator to match requests with resources available in the m-marketplace. Both requests and resources are expressed as semantically annotated descriptions, so that a semantic distance can be computed as part of the ranking function, to choose most promising resources for a given request. The rest of the paper is organized as follows: in Section II we briefly introduce the state of the art on service discovery in wireless ad-hoc contexts; in Section III we explain our approach to the problem; in Section IV a case study is presented to clarify the proposed framework; in Section V we comment on related work and in the last SectionVI we expose conclusion and future work. II. STATE OF THE ART Usually, resource discovery protocols involve a requester, a lookup or directory server and finally a resource provider. Most common SDPs, Service Location Protocol (SLP), Jini, UPnP (Universal Plug aNd Play), Salutation or UDDI (Universal Description Discovery and Integration) among others, include registration and lookup of resources as well as matching mechanisms [1]. All these systems generally work in a similar manner. Basically a client issues a query to a directory server or to a specific resource provider. The request may explicitly contain a resource name with one or more attributes. The lookup server –or directly the resource provider– attempts to match the query pattern with resource descriptions stored in its database, then it replies to the client with discovered resources identification and location [2]. These discovery architectures are based on some common assumptions about network infrastructure under the application layer in the protocol stack. In particular, current SDPs usually require a continuous and robust network connectivity, which may not be the case in wireless contexts, and especially in the ad-hoc ones. In fact in such environments, network consistence varies continuously and temporary disconnections occur fre- quently, bringing to a substantial decrease to traditional SDP performances [3]. Actually there are several issues that restrain the expansion of advanced wireless applications. Among them, the variabil- ity of scenarios. An ad-hoc environment is based on short range, low power technologies like Bluetooth [4], which grant the peer to peer interaction among hosts. In such a mobile infrastructure there could be one or more devices providing and using resources but, as a MANET is a very unpredictable environment, a flexible resource search system is needed to overcome difficulties due to the host mobility. Furthermore, existing mobile resource discovery methods use a simple string-matching, which is largely inefficient in advanced sce- narios as the ones related to electronic commerce. In fact in