Doychev, DoychinDoychinDoychevRafter, RachaelRachaelRafterLawlor, AonghusAonghusLawlorSmyth, BarryBarrySmyth2017-04-202017-04-202015 Sprin2015-06-03http://hdl.handle.net/10197/8442UMAP 2015: 23rd International Conference on User Modeling, Adaptation, and Personalization, Dublin, Ireland, 29 June - 03 July 2017In this paper we share our experiences of working with a real-time news recommendation framework with real-world user and news data. We discuss the challenges faced while working in such a noisy but uniquely real-world context. Specifically, we focus on an initial evaluation of a 12 different news recommendation algorithms across 7 different German news sites, including general news, sports, business, and technology related news sites. We compare the performance of these algorithms, paying particular attention to their relative click-through rates and how this can vary with time of day and news domain.enThe final publication is available at Springer via http://dx.doi.org/10.1007/978-3-319-20267-9_28 .Recommender systemsNew recommender systemsReal-time recommendation frameworksAdapting/contextualizing recommendationsConference Publication33734210.1007/978-3-319-20267-9_282016-01-19https://creativecommons.org/licenses/by-nc-nd/3.0/ie/