Evaluating Music Discovery Tools on Spotify: the Role of User Preference Structure

This study aim to explore the subjective evaluations and objective effectiveness of different music recommendation tools offered by Spotify, a music social software, and how does this result relate to the preference structure of users. We employ a mixed methods research in this study, attempt to assess the effectiveness of music recommendation tools more precisely in multi-dimensions. A quantitative research was conducted to evaluate the objective effectiveness and subjective evaluations: 1. The participants’ subjective evaluations was measured by a small-scale empirical experiment, setting the four music recommendation tools offered by Spotify as independent variable, and the preference structure of users as intermediary variable, and the subjective evaluations collected in the questionnaire as dependent variable. 2. The objective effectiveness was calculated by the proportions of tracks’ sets generated in the experiment. And we conducted an unstandardized in-depth interview after the experiment as a qualitative research method, making some triangulations, supplements and explanations. Some major findings are as follows. First, different music recommendation tools do have a difference of effectiveness. Second, the subjective evaluations given by users are not exactly consistent with the objective effectiveness of different music recommendation tools. Furthermore, the users’ preference structure do have an impact on the effectiveness of music recommendation tools.