Applying Bayesian Network and Association Rule Analysis for Product Recommendation

In recent years, there have been more and more enterprises using Web sites for marketing of various products or services; Internet thus allows customers shopping or searching for information online at any time and any location. If it is possible to recommend products to customers’ liking at the time they are visiting the specific web site, it would reduce the hassle customers experience in searching for products from a large information base. Moreover, it would increase the sales of advertised products. This research tries to make use of model-based Bayesian network to construct a product recommendation system. This research also uses association rule analysis to assist constructing the Bayesian network structure. This research expects to incorporate the two techniques to improve the shortcoming of a single technique. Especially, visualization of Bayesian network has also being used so that enterprises can easily observe the shopping behaviors of customers towards various products. This research integrates the inference results obtained from individual customers and the entire group in hopes of making surprisingly potential recommendation based on the preferences of the entire group in addition to recommending associated products favored by the given individual.