Norm based service selection

Abstract

Distributed computing paradigms are increasingly moving towards collections of interoperating Web services. To facilitate this interoperation, dynamic discovery and selection of services is required. Existing distributed solutions for the dynamic discovery of services primarily focus on the deployment of directory, broker and matchmaking intermediaries, requiring third party participation and additional infrastructure costs.

The selection of Web services by autonomous actors has become a well-developed area of research. Service-oriented architectures can now provide for complex interactions described by semantically rich process models, thereby enabling consumption by autonomous agents. With distributed agent-based architectures becoming common, academics are increasingly looking towards norm-based approaches to offer flexible control of interacting agents.

Current semantically aware service selection methods rely on matching inputs and outputs provided by services profile models. This approach typically fails to allow actors to differentiate between services where the profile models may match, but the process models differ.

In this research, the question is asked: How can an actor with a set of known normative beliefs use these beliefs to aid service selection where IOPE matching typically falls short?

The following have been created: a model, a language and a module for the norm-based scoring of process definitions. In doing so it is shown that social norms can be used by actors to reason over the potential cost of any interaction and that this metric can provide useful context when selecting partners where the services' basic inputs and outputs may match, but process model specifications may not.

More information

Identifiers

Catalogue record

Export record