Modeling Web Site Personalization Strategies

Personalization is a key factor for differentiating services and retaining customers in World Wide Web sites. On the other hand, designing and implementing an efficient personalization strategy is still a challenge, because of the complexity of the techniques used and the variety of sites and customers, which are always evolving. This paper presents a functional model of personalization strategies that allows not only a simple and concise specification of those strategies, but also their simulation and validation. We demonstrate our model through e-Personal, a framework for estimating the effectiveness of personalization strategies. The framework guides the user through the process of specifying a strategy and estimates its impact based on previous interactions of customers with the site. It is based on our functional model and we illustrate its utilization for designing personalization strategies for a web portal. Our experiments are based on actual logs and show that the proposed framework enhances significantly the personalization process, indicating the goodness of the strategy design, the reliability of input data, and the impact of implementation decisions on the effectiveness of personalized sites.