The Institute for Information Law (IViR) at the University of Amsterdam is one of the academic partners in the Horizon Europe Consortium Open Music Europe
, consisting of 15 partners in 11 countries. Open Music Europe strives to fill the data gaps identified by 150 music industry stakeholders. We create usable indicators that allow music researchers, music businesses and cultural policymakers to use our data as evidence in more informed institutional, business, or public policies. In the context of this project, the IViR
interdisciplinary research team is looking for a postdoctoral researcher interested in music and statistical innovation. Background
Recorded music has been sold and streamed on global platforms for more than a decade now, which makes the sector one of the most data-rich business environments in the world. Technically, there is a trace of almost every song listened to in any city in the world. In Europe, music stakeholders are small; therefore, they only see a very small part of this global music picture. If there is an increased demand for their catalogue, they do not know if there are more music listeners, their genre is more sought-after, they get traction in better-remunerated territories, currency exchange rates work in their favour, or they reach a new accrual period. We are developing quantitative tools to generalize this information and provide better data and metadata for an economically and socially more sustainable music ecosystem for Europe. What are you going to do?
We are looking for a candidate familiar with dynamic sampling techniques and can apply the law of big numbers in practice. You will help design an algorithm that will allow us to select a basket of songs to view if the observed mean, median, standard deviation and other statistical properties of their play count and revenue are starting to correlate in the baskets of each other. When they show the same statistical picture, we can be sure that we are selecting songs randomly, without any inherent bias in the selection algorithm, such as the digit distributions of an alphanumeric identifier.
The Successful candidate will work on the challenge to select "typical songs" that well represent the sales prospects, popularity, and use of songs played in a given genre, or country (territory) in a given royalty accounting period (a month). We can access detailed data about where and how many times a song was played and what the financial remuneration was. However, we need to learn how to select a song randomly because we are unfamiliar with the (statistical) population of the songs available on a particular streaming platform or in a country. We can query databases based on the International Standard Recording Code (a unique identifier for recordings) or via a unique identifier given by MusicBrainz, Spotify, YouTube, but we do not yet know how to select randomly. Our problem is similar to creating a representative "basket" of stocks or bonds on the financial markets that represent the market sentiment in a stock or bond market index or selecting in practice participants in a repeating pseudo-panel social sciences survey. In addition to this core task and depening on progress, we will discuss futher work within this Open Music Europe project that the candidate will do.