| Abstrakt: |
The aim of this paper is to provide a general ontology that allows the specification of trust requirements in the Semantic Web Services environment. Both client and Web Service can semantically describe their trust policies in two directions: first, each can expose their own guarantees to the environment, such as, security certification, execution parameters etc.; secondly, each can declare their trust preferences about other communication partners, by selecting (or creating) ‘trust match criteria'. A reasoning module can evaluate trust promises and chosen criteria, in order to select a set of Web Services that fit with all trust requirements. We see the trust-based selection problem of Semantic Web Services as a classification task. The class of selected Semantic Web Services (SWSs) will represent the set of all SWSs that fit both client and Web Service exposed trust requirements. We strongly believe that trust perception changes in different contexts, and strictly depends on the goal that the requester would like to achieve. For this reason, in our ontology we emphasize first class entities "goal", "Web Service" and "user", and the relations occurring among them. Our approach implies a centralized trust-based broker, i.e. an agent able to reason on trust requirements and to mediate between goal and Web Service semantic descriptions. We adopt IRS-III as our prototypical trust-based broker. [ABSTRACT FROM AUTHOR] |
