Composition of Trustworthy Web Services *Karl Quinn, Declan O’ Sullivan, Dave Lewis, Vincent Wade. Knowledge and Data Engineering Group, Trinity, College Dublin, Dublin 2, Ireland. Tel: 00 353 1 608 1355 Email: firstname.lastname@cs.tcd.ie Abstract: An ontology based trust model that can be used to semantically annotate web services was designed and developed to facilitate the computation of trust values for service selection and composition. This paper describes the research areas and presents an analysis of the salient issues experienced throughout a three month R&D project. Acknowledgement: This work was carried out in the Ericsson Ireland Research Department as part of an Internship by the lead Author during 2004. The project was a joint effort between Ericsson Ireland and the Knowledge & Data Engineering Group (KDEG), Trinity College Dublin. The lead author is funded through an Irish Research Council for Science, Engineering, and Technology (IRCSET) scholarship. I. Introduction Definitions of trust generally use synonyms or trust inspiring terms. “Belief” [i], “Credibility or Reliability” [ii], “Confidence or Faith” [iii], “Reputation” [iv], and “Competence and Honesty” [v] have all been used in this way. Definitions generally try to convey that trust has a specific, quantitative, and directed (i.e. A to B, not B to A) value. Increasingly the definition of trust values and their calculation are seen as important elements of an overall security framework. Determination of an end-to-end trust value for a particular service can be brought about by reasoning over trust metadata associated with each of the individual service components. Web services will benefit from the use of trust metadata and management as it can aid in the automatic discovery or composition of trustworthy web services. In parallel, a major direction of web service research is towards service collaboration and semantic annotation through ontologies [vi]. Ontology provides a means to describe and define terms, concepts, and relationships specific to a knowledge domain such as trust and services. The development of technologies for the Semantic Web [vii] has produced ontology languages such as the W3C’s OWL (Ontology Web Language) [viii] which can be reasoned over at runtime. It is argued that trust and service metadata described with reference to such ontologies provide for improved reasoning because of its semantic basis. Trust management research is also exploring models for managing trust on an internet scale outside of the more traditional centralized systems. Furthermore trust management research has also exploring the use of policy based management [ix, x, xi, xii]. Given the above directions, our research focuses on combining ontology and policy based approaches for trust management. This involves the annotation of web services with trust metadata with reference to an ontology for trust and services. In addition users will be enabled to specify policies that determine how the management system should reason over the trust metadata for trustworthy service selection and composition. Karl Quinn Digitally signed by Karl Quinn DN: CN = Karl Quinn, C = IE, O = Trinity College Dublin, OU = Knowledge & Data Engineering Group Reason: I am the author of this document Date: 2004.10.05 16:32:25 +01'00'