Bringing communities of buyers and sellers together in the arena of electronic commerce stimulates three major potentials: the building of trust, the collection and effective use of community knowledge and the economic impacts of accumulated buying power. In this context, we introduce the concept of Virtual Communities of Transaction and review important personalization approaches which we may utilize in their design: collaborative filtering, data mining, and techniques to optimize the user interface and the underlying product offerings. The key contributions of this paper are the elaboration of Virtual Communities, the presentation of a categorization scheme for different types of communities, the identification of classes of member profiles, and the innovative concept of community products. We conclude with the case of the Amazon.com Recommendation Center to illustrate key design ideas and discuss an evolutionary application, the Participatory Product Catalog.