Sommaire:
One of the ultimate goals of the web service technology is to enable rapid low-cost development and easy composition of distributed applications, a goal that has a long history strewn with only partial successes. The research problems involved by service composition are varied in nature and depends mainly on the kind of the composition process, i.e., manual v.s. automatic, on the model used to describe the services and on the issues related to composition. A line of demarcation between existing works in this area lies in the nature of the composition process: manual v.s. automatic. The first category of work deals generally with low-level programming details and implementation issues (e.g., WS-BPEL) while automatic service composition focuses on different issues such as composition verification, planning or synthesis. In this talk, we will review recent research works and challenges related to formal verification and synthesis of service composition. Then we consider more particularly the composition synthesis problem, i.e., how to generate automatically a new target service by reusing some existing ones, and discuss the associated decidability problem in presence of both bounded and unbounded number of instances.