A group-based inference approach to customized marketing on the Web integrating clustering and association rules techniques

Because of the intelligent computing specialty of the World Wide Web, extensive customized marketing can be executed at much lower cost and has become an emerging research issue. Therefore, the first purpose of this paper is to propose a system framework to serve as a foundation for developing a customized marketing system on the Web according to the discussions on data sources, data categories, and inference foundations. Most previous studies used induction-learning techniques to perform individual-based inference for customized marketing. However, it not only costs more to learn the personal preferences, but also some difficulties occur from using induction-learning techniques. The second purpose of this paper is to solve these problems. A group-based approach that integrates clustering and association rules is proposed. We conducted a field study to collect data to demonstrate the proposed group-based inference approach and evaluate its performance. The results reveal that this integrated approach can learn both more detailed and precise rules.