The Workshop of the Recommender Systems Challenge of ACM RecSys 2012 (RecSysChallenge 2012) will take place as a full-day workshop on Thursday September 13th and will run as an interactive and hands-on workshop during which participants will be guided through the whole process of participating in such a challenge. Taking as examples the datasets that drove the contest of this year’s challenge (with data from the Moviepilot and Mendeley platforms), we want to present, discuss and reflect on all the important steps of developing, testing and deploying recommendation algorithms over a real-life dataset – starting from acquiring and processing the dataset, and ending to performing an evaluation that will produce results comparable and reproducable by others.
RecSysChallenge 2012 generally focuses on how researchers and developers may use datasets from existing real-life systems in order to develop, simulate, benchmark and deploy their recommendation algorithms and services. It invited interested research and development groups to work with two real datasets in order to experiment with different types of recommendation approaches: a dataset from the Moviepilot.com platform was made available for the Benchmarking Track on Context-Aware Movie Recommendation (CAMRa), whereas a dataset from the Mendeley.com platform supported the Exploratory Track on Scientific Paper Recommendation (ScienceRec).