Using Ontology-based BDI Agent to Dynamically Customize Workflow and Bind Semantic Web Service

Abstract

As the Web gradually evolves into the semantic web, the World Wide Web consortium (W3C) recommends that web ontology language (OWL) be used to encode semantic information content over the Web. Semantic web is an essential infrastructure to enhance Web to obtain better integration of information and intelligent use of web resources. Moreover, a web service is annotated by web ontology language for service (OWL-S) to form a semantic web service that, however, is a static description. The OWL-S based semantic web services thus are reactively invoked by users. How to dynamically coordinate, composite, or discover the services is an important issue.  We use agent to proactively interpret the static OWL-S description. And, the Belief-Desire-Intention (BDI) model is applied to develop BDI agent. This work thus proposes a ontology-based BDI agent architecture, in which a BDI agent dynamically generates customized workflow, and binds semantic web services. The architecture includes four parts: 1) Application Ontology, which is description of a specialized domain, 2) Operation ontology, which is description of BDI agent, 3) Ontology-based BDI agent engines, which interpret corresponding operational ontology to dynamically generate workflows, and 4) Java agent development environment extension (JadeX) platform that our architecture is based on. Through JadeX, our BDI agent can dynamically bind semantic web services according to customized workflows.