In IPTV systems, thousands of live TV channels and video contents are available to subscribers. To benefit from the rich set of contents, users need to be able to rapidly and easily find what they are actually interested in. The integration of a recommender system into the IPTV infrastructure improves the user experience by providing an effective way of browsing for interesting programs and movies. But to improve the quality of recommendation, a clear understanding of access pattern to the items is necessary. However, for security reasons, in IPTV systems this kind of information is not publicly available. In this paper, we try to generate a so-called artificial dataset based on a model which mimics the behavior of a typical IPTV user, and with the aid of MovieLens dataset. We then show a typical application of the dataset in recommender systems.
