Proceedings of the 12th ACM Conference on Recommender Systems (RecSys 2018)
Original Kurzfassung:
This tutorial introduces multimedia recommender systems (MMRS),
in particular, recommender systems that leverage multimedia content to recommend different media types. In contrast to the still
most frequently adopted collaborative filtering approaches, we
focus on content-based MMRS and on hybrids of collaborative
filtering and content-based filtering. The target recommendation
domains of the tutorial are movies, music and images. We present
state-of-the-art approaches for multimedia feature extraction (text,
audio, visual), including deep learning methods, and recommendation approaches tailored to the multimedia domain. Furthermore,
by introducing common evaluation techniques, pointing to publicly
available datasets specific to the multimedia domain, and discussing
the grand challenges in MMRS research, this tutorial provides the
audience with a profound introduction to MMRS and an inspiration
to conduct further research.