A dataset of multimedia material about classical music: PHENICX-SMM

Abstract

We present a freely available dataset of multimedia material that can be used to build enriched browsing and retrieval systems for music. It is one result of the EU-FP7 funded project “Performances as Highly Enriched aNd Interactive Concert experiences” (PHENICX) that aims at enhancing the listener experience when enjoying classical music. The presented PHENICX-SMM dataset includes in total more than 50,000 multimedia items (text, image, audio) about composers, performers, pieces, and instruments. In addition to presenting the dataset, we detail one possible use case, that of building a personalized music information system that suggests certain types and quantities of multimedia material, based on personality traits and musical experience of its users. We evaluate the system via a user study and show that people generally prefer the personalized results over non-personalized.