Artist Similarity with Graph Neural Networks

TL;DR

This paper introduces a novel graph neural network approach for artist similarity that combines artist connection topology with content features, validated on a large new dataset and outperforming existing methods.

Contribution

The paper presents a hybrid GNN model for artist similarity, introduces the OLGA dataset, and demonstrates scalability and superior performance over current methods.

Findings

The GNN approach outperforms state-of-the-art methods.

OLGA is the largest artist similarity dataset with content features.

The method scales to larger proprietary datasets.

Abstract

Artist similarity plays an important role in organizing, understanding, and subsequently, facilitating discovery in large collections of music. In this paper, we present a hybrid approach to computing similarity between artists using graph neural networks trained with triplet loss. The novelty of using a graph neural network architecture is to combine the topology of a graph of artist connections with content features to embed artists into a vector space that encodes similarity. To evaluate the proposed method, we compile the new OLGA dataset, which contains artist similarities from AllMusic, together with content features from AcousticBrainz. With 17,673 artists, this is the largest academic artist similarity dataset that includes content-based features to date. Moreover, we also showcase the scalability of our approach by experimenting with a much larger proprietary dataset. Results show the superiority of the proposed approach over current state-of-the-art methods for music similarity. Finally, we hope that the OLGA dataset will facilitate research on data-driven models for artist similarity.