A Domain Knowledge-based Personalized Comparison Shopping System: Design And Implementation

With the fast development of Electronic Business, the numbers of online stores and products are growing at a rapid pace. It becomes more and more difficult for a customer to find out the products which they interested. In order to find out the best products customers browse different online stores, collect information on merchants and products, compare products ,and then make decisions. But usually the product that he or she buys is not the best one.Comparison shopping system is a mediator system that automatically collects products descriptions from a number of online stores on a user's behavior.Most previous Comparison Shopping Systems were not effective. Manual wrapper construction is not scalable since the human developer must describe a new extraction rule for an information source that is newly created or structurally changed. Heuristic wrapper induction can generate wrappers automatically. However, this approach is not effective since the heuristics are mostly simple and naive, and the system can extract only a limited number of features such as name and price of a product.In order to resolve these problems, this paper proposes a domain knowledge-based wrapper induction (DKWI) approach.. This approach applies the domain knowledge to generate the wrappers for online stores, and then use these wrappers to extract information of products. DKWI is expected to extract not only the name and the price information of a product, but also more other features information such as the author name, publish house and ISBN number of the book .We have built a prototype comparison shopping system named eShopper with the DKWI approach. The eShopper System automatically generates a wrapper for each online store. When received a comparison request from a customer, eShopper will extract information of product descriptions from a number of online stores by applying corresponding wrappers, then integrate these product descriptions in a structured format and return a list of sorted products descriptions.