The chapter presents the SWAPTeam participation at the ECML/PKDD 2011-Discovery Challenge for the task on the cold start problem focused on making recommendations for new video lectures. The developed solution uses a content-based approach because it is less sensitive to the cold start problem that is commonly associated with pure collaborative filtering recommenders. The Challenge organizers encouraged solutions that can actually affect VideoLecture.net, thus the proposed integration strategy is the hybridization by switching. In addition, the surrounding idea for the proposed solution is that providing recommendations about cold items remains a chancy task, thus a computational resource curtailment for such task is a reasonable strategy to control performance trade-off of a day-to-day running system. The main contribution concerns about the compromise between recommendation accuracy and scalability performance of proposed approach.
CITATION STYLE
Iaquinta, L., Semeraro, G., & Lops, P. (2013). Cold start problem: A lightweight approach. Studies in Computational Intelligence, 439, 15–32. https://doi.org/10.1007/978-3-642-31546-6_2
Mendeley helps you to discover research relevant for your work.