[Zurück]

P. Filzmoser, N. Mumic, R. Kostadinova:
"Fraud detection in the digital music industry";
Hauptvortrag: Benford's Law Conference, Stresa (eingeladen); 10.07.2019 - 12.07.2019.

The digital music industry is getting more and more important, and this implies that data from music sales also get bigger. It is
practically impossible for the artists to check for the correctness of their sales figures, and thus there is an urgent need for tools to detect eventual frauds. Such frauds could originate from different platforms, but it would be unlikely that fraud happens at the same time on all platforms. Thus the idea is to design a fraud detection method which is making use of the (log-)ratios between the
values of the different platforms. Potential fraud would happen if the log-ratios are large.

Compositional data, robust filtering, outlier detection

https://publik.tuwien.ac.at/files/publik_282799.pdf