Flat and hierarchical user profile clustering in an e-commerce recommender system

Recommender systems are more and more used in different domains of computer science. The collaborative filtering remains a highly prized recommendation technique used by the e-services on the internet. This technique is mainly based on deducing a part of the user interests from the preferences of other users with similar profiles. Among the different approaches, the clustering technique is used to implement collaborative filtering. We propose in this work a comparison between hierarchical and flat user profile clustering based on a case study. The proposed approach is implemented basing on a dataset of user profiles in an e-commerce context.