Adaptive Middle Agent for Service Matching in the Semantic Web: A Quantitative Approach

With the advent of the Web services and the need for a Semantic Web, the agent technology is (in my view) finally becoming a viable solution to many real world problems. Effective service matching is key to the success of agent systems, but existing service matching methods mostly consider service descriptions as the only factor and many important issues have been largely overlooked. In the real world applications, agents with identical service descriptions may differ dramatically in performance levels; an agent may have strong and weak areas in its service offerings; and the distribution of services may provide helpful hints on which matches are more likely to be better than the others. Moreover, an agent's capability may change over time. These are very important issues and if left unaddressed, may become real problems in the real world applications.

In our approach presented in this dissertation, the middle agent establishes and refines an agent's capability model based on the domain ontology and through the interactions with the agents. In this framework, an agent's performance history is considered as an integral part of the agent's capability model and the agent's strong and weak areas can also be revealed. Moreover, the dynamically captured and updated service distribution in the service domain is considered as an important factor in service matching. Service matching here is carried out in two steps. In the first step, candidates are selected through the semantic service description matching. In the second step, the performance rating of each candidate with respect to the specific request is estimated based on the agent's capability model, and the candidates with the highest estimated performance ratings will be selected.

The major advantages over existing methods are the establishment and refinement of an agent's capability model and the use of such a model in service matching. A prototype system and an evaluation framework have been implemented to evaluate the ideas discussed in this work. The statistics collected from the experiment shows a significant improvement over typical service matching methods in terms of the accuracy in selecting the best service provider(s) for each request.