The book chapter focuses on recommender systems based on analyzing web browsing logs and clustering. The positive part for me is that it introduces into concepts of collaborative filtering, recommending shortcuts and association rules. It explains that collaborative filtering is less appropriate as we are dealing with non-intrusive measurement and no rating. After this introduction, the paper sketches hybrid methods and goes into detail into a system developed based on clustering and generating navigational patters. Here the author lost me, because of focus on web sessions and page visits, which is distant from the testing recommender system we are interested in. Moreover, the paper contains spelling mistakes and due to engineering focus goes further away from scientific writing I am used to. It also focuses on the difficulties of using web log data: Incomplete information, incorrect information and the persistence problem wen new web pages arrive.

10.2495/1-84564-152-3/05

O.R. Zaïane, Recommender systems for e-learning: towards non-intrusive web mining, WIT Transactions on State of the Art in Science and Engineering, Vol 2, 2006, pp. 87-96

Eligius Hendrix

University of Malaga