Aggregation and maintenance for database mining

Abstract

Database mining must capture the dynamics of data in the real-world. In other words, mining models would be able to maintain all changed rules when a database is updated. This paper presents a new model of aggregating association rules aimed at not only maintaining association rules in dynamical databases, but also aggregating association rules from different data sources. Indeed, this aggregation of association rules is useful for making decisions. Our experiments show that this model is efficient to aggregate rules from both of dynamical databases and different data sources.