Visualization of the artist relations using twitter user profiles

Abstract

We propose a new music artist recommendation algorithm using Twitter profile texts. Today, music recommendation is provided in many music streaming services. In this paper, we propose a new recommendation algorithm for this music recommendation task. Our idea is to use Twitter profile texts to find appropriate artist names to recommend. We obtained word embedding vectors for each artist name by applying word2vec algorithm to the corpus obtained by collecting such user profile texts, resulting in vectors that reflect artist co-occurrence in the profile texts.