Brokerage for mathematical services in MONET

As a result of the rapid increase in the number and range of services deployed on the Internet, it is becoming apparent that finding services, choosing between them and working out how to combine them is a desirable, but also very challenging problem, which has brought into existence numerous languages for describing the myriad aspects of web services. We describe here an approach to service brokerage being explored as part of the MONET project [12], whose larger aim is to demonstrate the applicability of the semantic web to the domain of mathematical software. The enabling technologies originate from artificial intelligence, with the objective of matching the characteristics of a problem to the advertised capabilities of available mathematical web services and then invoking the chosen web services through standardised mechanisms. The resulting framework aims to be a powerful, flexible, distributed and dynamic way of putting state-of-the-art algorithms at the disposal of users anywhere in the world. The scope of this paper is limited to service brokerage in the framework of the semantic web in which we use and redefine and extend features previously described in the RETSINA and InfoSleuth architectures by the incorporation of FIPA-ACL messaging, the ACL Infrastructure and the Mathematical Service Description Language (MSDL) which was developed in the MONET project to advertise capabilities of mathematical services. In addition we define a new layer, whose task is classification, thereby easing and distributing the problem of searching for mathematical services.