A New Architecture for Personalization Engines: An Open Source Approach

There are many benefits in having personalized engines on a Web site. Several types of architecture have evolved to support this information system. Even though, a recent survey has shown that online shopping behavior is not altered by installing these systems on a Web site, enormous money is being spent on their use. These commercial personalization engines are very expensive to buy and are proprietary in nature. One alternative advocated in this article is to use a new architecture that follows an open source philosophy and uses a situation questions, problem questions, implication questions, need-payoff questions (SPIN) based question-answering strategy to interact with visitors. An implementation (called JESPER, a Jess-enabled personalization system) of such an architecture using JESS (Java-based expert system shell) is also presented. Our experience shows that personalization engines built this way for a Web site can be quite cheap and rigorous.