A framework for genetic-immune integration in data mining

In this paper, a novel approach is presented to solve the problems of dynamic data mining(DM) such as low effectiveness, high randomness and hard implementation. With the extension of the concept data mining process, the evolutionary and immune characteristics in dynamic mining are first illustrated respectively, based on the facts of the relationship between data mining tasks and the global optimization and of the comparison between the dynamic mining process and biological immune one. Then, a framework for genetic-immune integration is designed, where a new concept of fuzzy tracking is proposed and in turn a robust coordinator is constructed to ensure the effectiveness and robustness of the framework. Additionally, a new algorithm based on the framework for deep Web mining is described in detail and experiments are finished to prove the new approach correct and effective. Finally the work and proposals for the future are concluded.