Patricia Lynch Carbone and Larry Kershberg
Recent years have seen the increasing development of knowledge discovery or database mining systems that combine database management technology with machine learning techniques and algorithms to perform the analysis of data. Many of these systems use passive database management .systems to hold the data rather than active databases. However, applications such as battlemanagement situations or stock market trading would benefit from the use of an active database management system since the data is being constantly updated and other actions should be triggered based on database events. In this paper, we present a general description of an active knowledge mining system that combines an active database with machine learning operators. We introduce the notion of an intelligent mediator that contains knowledge about both the database and the capabilities of the learning operators. The mediator chooses which of the operators to use to achieve a learning goal, then determines how the discovered knowledge should be used and where it should be stored and maintained.