As the Semantic Web develops it is anticipated that many different
kinds of services will be exposed as Web Services. The discovery,
invocation, and monitoring of these Semantic Web Services will be much
more involved, and powerful, than is achievable with today's handraulic
It is my contention that the current brokering facilities on the Semantic Web are inadequate, and will require richer representations of the processes involved, as well as an ability to incorporate uncertainty in reasoning about service matches and service composition. I examine some of the issues involved, and suggest possible solutions, which will be investigated in my subsequent PhD.