Article titled “Offline Evaluation for Recommender Systems” has been published in Journal of the Japanese Society for Artificial Intelligence.
This is what I wrote while working in GroupLens Research in University of Minnesota. This covers most of the existing evaluation metrics for recommender systems. Actually it covers not only evaluation metrics related to recommendation correctness like precision, recall and F-measure but also those related to users’ discovery like novelty, serendipity and diversity. I recommend this article not only for recommender systems researcher but also for researchers on information retrieval, pattern recognition and machine learning. Please check it.
Journal of the Japanese Society for Artificial Intelligence, Vol. 29, No. 6, pp. 658-689, 2014.
English slide version [PDF]