Rule-based agents for the semantic web

摘要

Artificial agents, subsuming both robots and software agents, represent a new paradigm in software engineering and artificial intelligence. Depending on the technologies used in their implementation, they may exhibit various skills; in particular, they may act more or less autonomously, they may be able to learn and to adapt to a changing environment, and they may be able to pursue their goals pro-actively. An artificial agent is called rule-based, if its behaviour and/or its knowledge is expressed by means of rules. In this paper, we discuss a general architecture for rule-based agents and how it can be realized with the help of semantic web languages. We also show how such agents can go live on the web by presenting an implementation in Mandarax, a Java rule platform. The concept and implementation are complemented by a running example, the portfolio agent.