Grant: Discovery Grants Program - Individual

This proposition program covers and focuses manly on protocol search and queries.Current tools support search based on the service interface which is adequate for searching stateless services. However, the interface is absolutely not enough to be representative for stateful services which are characterized by their behavioral models. Conceptually, the behavioral model of stateful service is a protocol that can be expressed in formalisms such as which finite state machines, petri nets, activity diagrams, $\pi$-calculus, etc.Given a protocol, some researchers have already investigated how to find out services that exactly match that protocol with respect to a kind of bisimulation relation. Such relation is very strong and restrictive. Consequently, there is very little chance that a match occurs. Such data (which are protocols), like images and videos, cannot be represented in canonical format that is meaningfully searchable on the basis of database queries seeking an exact matches. To overcome this limitation, we propose to investigate how similarity metrics can be computed for each service protocol of the registry according to a given protocol of the query. A designer would be invited to closely consider the highly similar protocols, which are the result of the query, as they may fulfill his needs. In the case of protocol search, it would be very improbable that an exact match occurs while the registry may contain relevant protocol for the designer's needs. Avoiding the situation of miss, means more cost and effort savings, and more business opportunity and partnership in the context of SOA. The preliminary study reveals that this proposal will fill a gap at least in service discovery support. Even though this proposal takes SOA as an application domain, its fundamental results are general and applicable whenever a protocol similarity search is relevant such as in component based approaches.