Policy-Driven and Multi-agent Impact Identification Model for Component-Based System

In this paper, we propose a policy driven and multi-agent based model to identify the impacts of the new released components on the existed component-based system, which will be used to select test cases for regression test. In our model, policy, expressed in XML, defines the association between certain types of documents of components and corresponding analysis algorithms. Based on the policy, our model can automatically adopt suitable algorithms to analyze the impacts. The whole procedure is driven by three agents: parse agent, query agent, and dispatch and execution agent. Parse agent will parse the input XML files for query agent to retrieve policies for analysis procedure. Dispatch and execution agent will adopt the corresponding algorithm to analyze the modifications from the new version of components to the old ones according to the policies. Such a policy-driven and multi-agent based model can be self-adapted for automatic selection of suitable analysis algorithms to do the impact identification for component-based system.