UbiquiTo-S: A Preliminary Step Toward Semantic Adaptive Web Services

UbiquiTo-S: A Preliminary Step Toward Semantic Adaptive Web Services Francesca Carmagnola, Federica Cena, Cristina Gena, and Ilaria Torre Dipartimento di Informatica, Università di Torino, Corso Svizzera 185, Torino, Italy {carmagnola, cena, cgena, torre}@di.unito.it

Abstract. In this paper we describe an approach to design an adaptive system as a Semantic Web Service. We focus on how adaptive systems can provide adaptive services through web service technologies. In particular, we concentrate on adding semantic information to enrich the service discovery phase. We present a recommender system, UbiquiTO-S, which exploits the technology of Web Services (WS) and Semantic Web to allow software agents to discover its services and use its adaptive services.

1 Introduction and Basic Principles The past years have shown an enormous increase of efforts in standardization and development of interoperable Web technologies. In particular, with the number and diversity of Web Services (WSs) and Semantic WSs expected to grow, a recent trend is targeting the vision of offering adequate techniques for user-centric and preferencebased services discovery and selection technologies that support personalization [2]. Adaptive techniques may be employed in the process of interaction among WSs. As underlined in [3], this issue leads to a number of research questions: How can Semantic WSs be used in adaptive environment? How can Adaptive Systems and WSs be joined to meet user needs? The most common approach is to add personalisation to the different stages of the process of interaction among WSs: discovery, querying, and execution [7,2]. In particular, a service can be searched according to user personal profile, which typically includes her individual preferences together with technical constraints on terminal and environment. A like-minded approach can be found in [3], which furthermore introduces the composition of Semantic WS by a domain expert, and the adaptive integration service in response to user needs using ontologies in order to capture and exchange models of the real world and making them available to automated agents. Our approach moves from the same considerations, but the perspective is different: the focus is on Adaptive Systems that provide adaptive services using Semantic WS technologies. A similar approach is led by [6] that developed a mobile tourist application that offers information and services tailored to the context and the user plans. The platform is based on WS technologies combined with Semantic Web technologies, it is open and allows third parties to integrate new services and information into the platform. Similarly, in our approach we focus on the adaptation of an application that is designed to work as a Semantic Adaptive WS. In particular, in this preliminary V. Wade, H. Ashman, and B. Smyth (Eds.): AH 2006, LNCS 4018, pp. 249 – 253, 2006. © Springer-Verlag Berlin Heidelberg 2006

250

F. Carmagnola et al.

work we concentrate on adding semantic information to enrich the service discovery phase. To extend WSDL1 descriptions with semantic information we use WSSP (Web Service Semantic Profile) since it is specifically designed for the description of service profile and thus for the discovery phase [4,5]. Moreover, to describe ontologies and rules referenced in WSSP we use OWL2 and RDF-RuleML3. In the next section we will present a usage scenario with the description of the corresponding platform to highlight that extending the expressive power of WSDL files is essential to implement a Semantic Adaptive WS. In particular, exploiting ontologies and semantic rules allow to express conditional inputs that are commonly necessary for user modeling and adaptation, in order to explain the reasons why such inputs are required and thus to explain the behaviour of the system. Moreover, semantic is essential to improve interoperability and to allow explanations, proof, user models sharing, etc. In the last part of the paper we will provide an example of a RDFRuleML adaptation rule linked by a WSSP [4,5] file in UbiquiTO-S, a semantic tourist recommender.

2 Usage Scenario and Platform Description USAGE SCENARIO Carlo is a business traveler. He arrives in Torino and looks for events in the late afternoon. Thus, he runs the browser of his SmathPhone GPS-equipped and queries his preferred matchmaker4 in order to find out a WS that satisfies his needs. From the matchmaker search interface he selects the categories tourism and events. The matchmaker already knows some information about Carlo, such as his position provided by his GPS-receiver and his profession and age provided by Carlo when he registered to the service the first time he used it. The matchmaker discovers UbiquiTO-S, invokes its services and finally Carlo receives a list of events in the city of Torino that mostly correspond to his interests and propensity to spend. He is satisfied about the recommendations and also because both the discovery of UbiquiTO-S and the personalization of the answers were carried out automatically. PLATFORM DESCRIPTION The above example presents a scenario where several WSs provide services to endusers and to software agents. As typical, they describe their services in WSDL files and advertise them in public UDDI5 registries. UbiquiTO-S is a WS, and in particular a Semantic Adaptive WS that provides recommendations in the tourist domain. The matchmaker searches the UDDI registry and discovers several WSs (UbiquiTO-S is one of them) offering the information Carlo is interested in. The matchmaker evaluates the match between i) user requests and known user features and ii) the provided 1

http://www.w3.org/TR/wsdl http://www.w3.org/TR/owl-features/ 3 http://www.w3.org/2004/12/rules-ws/paper/93/ 4 A matchmaker is a search engine a user can delegate to find services. It works performing a match between the user request and the service description typically advertised on UDDI registries. 5 http://www.uddi.org/ 2

UbiquiTo-S: A Preliminary Step Toward Semantic Adaptive Web Services

251

outputs and WS requirements. The matchmaker decides to invoke UbiquiTO-S services since it is able to provide not only location based information, as done by other discovered WSs, but also information tailored to the provided user features. UbiquiTO-S is designed as a Semantic WS that semantically annotates required input, provided output and their relationships. In particular, UbiquiTO-S enriches the WSDL service description by means of a WSSP file that specifies restrictions and constraints for WSDL parameters of input and output. WSSP, which has been introduced and successfully exploited by [4,5], is a discovery mechanism that semantically enriches the UDDI’s search functionality and encodes semantic information with WSDL. It represents input and output semantics, as in the OWL-S profile6, by describing the service capabilities with semantic information expressed in some ontology language (for example OWL). In addition to OWL-S profile, WSSP gives the possibility to link rules that allow to better specifying restrictions and constraints to inputs and outputs (for example in RDF-RuleML). Many other standards have been proposed (e.g., WSDL-S7, WSML8, SWSL9) but the choice of using WSSP in our proposal depends on the fact that it is compatible with the service ontology of OWLS, and allows to use semantic rule languages, like RDF-RuleML, also in the description of service profile, and can be complementary used with the standard WSDL and UDDI structures. In particular, for the goal of adaptation, the possibility to define restrictions and constraints to input and output is quite interesting: 1. Restrictions to the required inputs can be specified by mapping the input parameter on ontologies. For example, an OWL ontology can be used to define the meaning of an input parameter (e.g., Profession), specifying its meaning and its relationships with other classes. 2. Constraints to input and output are expressed by means of semantic rules. Concerning the input, a rule can specify its allowed values, which in the above example are the subclasses of that particular class (e.g., Student, Manager). Concerning the output, rules are relevant to specify constraints that explain the reason why corresponding inputs are required. For example some rules exploited in the above scenario may specify that: • if the user profession is provided (input), suggestions regarding events (output) have a Confidence Value (CV) of 0.5; • if the user profession and age are provided (input), suggestions regarding events (output) have a CV of 0.7; Thus we can state that rules try to justify the required input by explaining that the corresponding output information will be more precise and tailored to the specific needs. This not only increases the control over the adaptive service and its scrutability, but also offers to the requestor user-tailored information in addition to the purposes the web service is designed for. In order to show the relations between WSSP and rules, Fig. 1 presents a part of the WSSP file that shows the output message and the URI reference to the rule that specifies the constraints to the output. Then Fig. 1 6

http://www. daml.org/services/owl-s/ http://www.w3.org/2005/04/FSWS/Submissions/17/WSDL-S.htm 8 http://www.wsmo.org/wsml/ 9 http://www.daml.org/services/swsl/ 7

252

F. Carmagnola et al.

presents a part of the RDF-RuleML rule referred to the specified URI in the WSSP file. In particular, as it can be seen, it expresses in RDF-RuleML the second rule of the above examples, i.e. “if the user profession and age are provided, suggestions have a CV of 0.7” WSSP constraints …. RDF-RuleMl

…. 0.7

Fig. 1. An example of WSSP constraints specified in a Rdf-RuleML rule

3 Conclusion and Future Works In this paper we presented a recommender system, UbiquiTO-S, which exploits the technology of Web Services and Semantic Web to allow software agents to discover its services and invoke its adaptive services. UbiquiTO-S has been developed as an extension of UbiquiTO [1], a mobile adaptive guide that provides personalized location-based tourist information. With respect to UbiquiTO, the main task is not changed, but its transformation into a WS and the choice to formally and semantically describe input requirements, outputs and especially their relationships introduces it in a cooperative environment for personalized services. With respect to common WSs, the advantage of this approach is that it allows a middle agent (e.g. the matchmaker in the scenario) to obtain services that i) better match the user request ii), fit specifically her features, iii) can be imported and processed to compose other services which take advantage of the personalization.

UbiquiTo-S: A Preliminary Step Toward Semantic Adaptive Web Services

253

Moreover, considering the user point of view, while UbiquiTO provides adapted information without explaining the reason why it needs some data and its adaptive behaviour it is not transparent, UbiquiTO-S, on the contrary, is more scrutable for both end users and software agents. As specified in the introduction, this is an ongoing work and this initial phase has been focused on the semantic discovery phase of the service, by adding semantic information to the service profile. The current work includes the definition of the WSDL file, the definition of the WSSP profile, and the definition of all the OWL ontologies and RDF-RuleML rules that are referred in WSSP to specify restrictions and constraints of the input and output messages. Our next step will deal with adding semantic to the whole process that includes service composition, execution and monitoring.

References 1. Amendola, I., Cena F., Console L., Crevola A., Gena C., Goy A., Modeo S., Perrero M., Torre I., Toso A.: UbiquiTO: A Multi-device Adaptive Guide. Proc. of Mobile HCI 2004, Lecture Notes in Computer Science, 3160 (2004) 409-414. 2. Balke, W., Wagner, M.: Towards personalized Selection of Web Services. In Proc. of the Int. World Wide Web Conf. (WWW), Budapest, Hungary, 2003. 3. De Bra, P., Aroyo, L., Chepegin, V.: The Next Big Thing: Adaptive Web-Based Systems. In Journal of Digital Information, Vol. 5 Issue 1 Article N.247, 2004-05-27. 4. T. Kawamura, J.-A. De Blasio, T. Hasegawa, M. Paolucci, K. Sycara, Preliminary Report of Public Experiment of Semantic Service Matchmaker with UDDI Business Registry, Lecture Notes in Computer Science, Volume 2910, Jan 2003, p 208– 224. 5. Kawamura T.; Hasegawa T.; Ohsuga A.; Paolucci M.; Sycara K.: Web services lookup: a matchmaker experiment. IT Professional Vol 7, Issue 2, Mar-Apr 05,,36-41. 6. Setten, M., Pokraev, S., Koolwaaij, J.: Context-aware recommendations in the mobile tourist application COMPASS. In Nejdl, W. & De Bra, P. (Eds.). AH 2004, 26-29 Aug 2004, Eindhoven, The Netherlands, LNCS 3137, Springer-Verlag, pp. 235-244. 7. Wagner, M., Balke, W., Hirschfeld, R., Kellerer, W.: A Roadmap to Advanced Personalization of Mobile Services. In Industrial Program of the 10th Int. Conference on Cooperative Information Systems (CoopIS 2002), Irvine, USA, 2002.