Music is omnipresent in modern societies. Consequently, attention to its selection, retrieval, and the maintenance of its collections is increasing. Using MusicSim, the authors introduce a utility to support these tasks. The authors start their paper with a concise, up-to-date overview on the topic.
Two main issues are addressed in this paper. First is the clustering of music through combining metadata and content-based (for example, audio) features; however, how this is implemented in MusicSim is not made clear. Second, the authors stress the need “to take the human into the loop.” As explained in the paper, this is done through an interactive card-sorting task, which provides relevant feedback to the system.
Although MusicSim’s rationale is good, the paper omits technical specifications and its implementation seems to lack advancement. Moreover, a thorough evaluation of the system is not described because it would be “beyond the focus of this paper.” So, at best, a true evaluation seems to be saved for another paper. Possibly, this paper could have illustrated MusicSim’s potential benefits, but, unfortunately, it also fails in this regard.