Authors: Nikos Manouselis, Constantina Costopoulou
Addresses: Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos Street, Athens 118 55, Greece. ' Informatics Laboratory, Agricultural University of Athens, 75 Iera Odos Street, Athens 118 55, Greece
Abstract: Recommender systems are deployed in electronic commerce (e-commerce) settings to help customers find products according to their preferences. Product recommendations may help buyers to save time by helping them choose from a variety of options. Recommendations that take into account the multiple attributes affecting a potential buyer|s decision can be particularly useful in the context of Business-to-Consumer (B2C) electronic markets (e-markets). Nevertheless, multiattribute recommender systems are usually more sophisticated than single-attribute ones, and their implementation may prove complex to e-market system developers. This paper presents the design, development and evaluation of marService, a product recommendation service that is based on Multi-Attribute Utility Theory (MAUT). This approach studies the application of marService for providing wine recommendations in an existing e-market and presents the results of a simulation experiment. Using an appropriate simulation environment, the evaluation of several design options for a set of algorithms for multiattribute utility recommendation has taken place, on two synthetic data sets for wine evaluations. Based on the experience from this experiment, some general suggestions that may prove useful to e-market developers wishing to implement a marService are also provided.
Keywords: recommender systems; e-markets; MAUT; multi-attribute utility theory; electronic commerce; e-commerce; electronic markets; business-to-consumer; B2C; multi-attribute recommendations; product recommendation; wine recommendations; simulation; wine evaluation.
International Journal of Computer Applications in Technology, 2008 Vol.33 No.2/3, pp.176 - 189
Published online: 10 Dec 2008 *Full-text access for editors Access for subscribers Purchase this article Comment on this article