Running in its third year at CLEF, NewsREEL challenged participantsto develop news recommendation algorithms and have them benchmarked inan online (Task 1) and offline setting (Task 2), respectively. This paper providesan overview of the NewsREEL scenario, outlines this year’s campaign, presentsresults of both tasks, and discusses the approaches of participating teams. Moreover,it overviews ideas on living lab evaluation that have been presented as partof a “New Ideas” track at the conference in Portugal. Presented results illustratepotentials for multi-dimensional evaluation of recommendation algorithms ina living lab and simulation based evaluation setting.@en

In real-world scenarios, recommenders face non-functional requirementsof technical nature and must handle dynamic data in the formof sequential streams. Evaluation of recommender systems musttake these issues into account in order to be maximally informative.In this paper, we present Idomaar—a framework that enables theefficient multi-dimensional benchmarking of recommender algorithms.Idomaar goes beyond current academic research practicesby creating a realistic evaluation environment and computing botheffectiveness and technical metrics for stream-based as well as setbasedevaluation. A scenario focussing on “research to prototypingto productization” cycle at a company illustrates Idomaar’s potential.We show that Idomaar simplifies testing with varying configurationsand supports flexible integration of different data.@en

Increasingly, educators make use of learning-by-doing approaches to teach studentsof STEM programmes the skills that they need to become successful incareers in research and development. However, we argue that the technicalchallenges addressed in these programmes are often too limited and thereforedo not support the students in gaining the more advanced skill sets required tothrive in our technology-oriented economy. We therefore suggest to incorporaterealistic and complex challenges that model real-world problems faced inindustrial settings. Focusing on the domain of recommender systems, we seepotentials in embedding recommender systems challenges to enhance studentlearning to teach students the skills required by modern data scientists.@en