Recommendation systems: A probabilistic analysis

A recommendation system tracks past actions of a group of users to make rec ommendations to individual members of the group. The growth of computer-med iated marketing and commerce has led to increased interest in such systems. We introduce a simple analytical framework for recommendation systems, inc luding a basis for defining the utility of such a system. We perform probab ilistic analyses of algorithms within this framework. These analyses yield insights into how much utility can be derived from knowledge of past user a ctions. (C) 2001 Academic Press.