Xiaohua Hu, Nick Cercone, and Jinshi Xie
Knowledge discovery in databases has attracted a lot of attention from the AI and databases community because of the huge information stored in the databases. There are a lot of algorithms developed to find rules from databases directly, but all these algorithms assume that the data and the data scheme are stable and most of the algorithm focus on discovering the regularities about the current data in the databases. In this paper we present a method which can learn rules from the :current data in the database to predict the data trend in the future. Our method combines the techniques of attribute-oriented induction, object-oriented databases and transition network. In our model, both the database contents and the database structure may evolve over the lifetime of a database.