Placing Music Artists and Songs in Time Using Editorial Metadata and Web Mining Techniques

Main test set

Our dataset contains 2868 Dutch and American song titles accompanied by artist names and release dates. Ground truth for both tasks (year of release, productivity estimation) can be generated in a straightforward manner. Our paper uses only the 15 most productive artists.

picture1 Download

Birth-death date test set

For those aiming at reproducing the birth/death date esimation results we provide the relevant dataset. It contains 663 artists with corresponding birthdates (prior to 1960) and deathdates. This data was gathered from MusicBrainz.

picture1 Download

Results Demonstration

Productivity profile

Year of Release

Latest article

01.08.2013 - Why time context?

The productive period of a music artist is important information that is typically highly correlated to his style, influences and similarities to other artists. Research has shown that the activity span is strongly associated with artist collaborations. As such, the productive years constitute a reliable, additional feature for various MIR tasks: listeners typically show a certain affection for music related to particular periods of their lives, and therefore time information could act as a basis for music recommendation. The practical applications of productivity profiles exceed the MIR domain. Productivity in absolute terms may be the most important factor for a comprehensive understanding of the creativity in music. Based on that, Kozbelt for example investigated the correlation between productivity and musical characteristics such as versatility and eminence. From a musicological perspective, a quantitative representation of productivity can offer valuable insights in music trends, significant musical events, significant social events, and the mutual influence that may exist between them.