A Grammatical Approach to User Model Resolution for Electronic

Electronic shopping is a rapidly expanding market and can reap beneets from the employment of user modeling techniques in order to personalize the shopping procedure for each customer. In this paper we present a methodology for user modeling that uses attribute grammars to extract a user representation from the navigation steps and actions the user takes inside the electronic shop. This method can be applied incrementally, demanding little space for user information, and avoiding the frequent recalculation of the entire user model. Shopping on the Internet constitutes a rapidly growing part of the worldwide market. E-commerce revenues, which are reported to be around $100 billion for 1999, are expected to grow as high as $1300 billion in 2003 3]. Research conducted in the US points out a 30% increase in Internet shopping from 1998 to 1999 and UK surveys demonstrate even higher increases, with an expected doubling of online sales between 1998 and 1999. European buyers in general are expected to be spending 17 billion Euro for online purchases by 2004 4]. Beneets of electronic shopping are twofold, applied both to the seller and the consumer. The most important of these is the potential for elimination of intermediaries between producer and consumer, which leads to the lowering of transaction costs and improvement of competition. A wider, possibly global market, is open to the seller, who can display the available products at a very low cost to a big audience. The lists and descriptions of available goods can be maintained with very little cost, especially compared to the costly maintenance procedure of printed catalogs. The customer, on the other hand, can save a signiicant amount of time and money with online shopping, not to mention the greatly improved convenience of the shopping experience, often a determining factor in consumer preferences. The information available to customers is ample and always up to date; it can be ltered by the potential buyer who wants to restrict the viewed data, in order to improve the selection and ordering process. Filtering of displayed information can also be performed by the sellers themselves, so that they can present the data in a manner and order more appealing to their audience. In order to select which data to display and how, the electronic shop software must be provided with clues about the user's preferences and buying habits. These clues can be incorporated under the general hood of …