The recent evolution of e-commerce and the astonishing growth of the Internet have increased the amount of information that scrupulous customers want to process before selecting items that meet their needs. Personalization has become an important strategy in Business to Consumer e-commerce, where knowledge about customers can be exploited in order to improve access to relevant products. This paper presents a machine learning-based approach to turn raw data about customers into knowledge about their interests. This knowledge is stored in personal profiles and is used to provide an intelligent search support.

Authors: Giovanni Semeraro, Paquale Lops, Marco Degemmis
Download: Link
Categories: H.3.3, H.3.5, i-KNOW 2003
Tags: Bayesian learning, intelligent search and retrieval, learning from labeled and unlabeled, personalization, user profiling