Web services are designed for composition and use by third parties through dynamic discovery. As such, the issue of interoperability between services is of great importance to ensure that the services can work together towards the overall application goals. In particular, the interaction protocols of a service need to be implemented and used properly so that the service composition can conduct itself in an orderly fashion. There have been significant research efforts in providing rich descriptions for Web services, which includes their behaviour properties. When describing the interaction process/protocols of a service, most of them adopt a procedural or programming style approach. We argue that this style of description for service interactions is not natural to publishing service behaviour properties from the viewpoint of facilitating third-party service composition and analysis. Especially when dealing with service with diverse behaviour, the limit of these procedural approaches become apparent. In this thesis, we introduce a lightweight, pattern/constraint-based declarative approach that better supports the specification and use of service interaction properties in the service description and composition process. This approach uses patterns to describe the interaction behaviour of a service as a set of constraints. As such, it supports the incremental description of a service's interaction behaviour from the service developer's perspective, and the easy understanding and analysis of the interaction properties from the service user's perspective. It has been incorporated into OWL-S for service developers to describe service interaction constraints. We also present a framework and the related tool support for monitoring and checking the conformance of the service's runtime interactions against its specified interaction properties, to test whether the service is used properly and whether the service fulfils its behavioural obligations. The tool involves interception of service interactions/messages, representation of interaction constraints using finite state automata and finite state machine, and conformance checking of service interactions against interaction constraints. As such, we provide a useful tool for validating the implementation and use of services regarding their interaction behaviour.