Abstract
Context-aware recommender systems are information filtering and decision support applications that generate recommendations by exploiting context-dependent user preference data, such as ratings augmented with the description of the contextual situation detected when the user experienced the item. In fact, many contextual factors (e.g., weather, season, mood or companion) may potentially affect the user’s experience of an item, but not all of them are equally important for the recommender system performance, or easy to be automatically acquired. Hence, it is important to identify and collect only those factors that truly affect the user preferences (ratings) and can improve the effectiveness of the recommendations computed by the recommender system. Extending our previous work, in this paper, we propose a novel method which adaptively elicits the most useful factors from the user upon rating an item. The proposed method deems a contextual factor as useful to be elicited when a user is rating an item, if it has an impact on the user’s predicted rating for that item. The results of our offline experiments, which we executed on travel-related rating datasets, show that the proposed method performs better than other state-of-the-art context selection methods. This paper is an extended and updated version of a conference paper titled ‘Contextual Information Elicitation in Travel Recommender Systems’ previously published in the proceedings of Information and Communication Technologies in Tourism 2016 Conference (ENTER 2016) held in Bilbao, Spain, February 2–5, 2016. © 2017, Springer-Verlag Berlin Heidelberg.