Polarized Score Distributions in Music Ratings and the Emergence of Popular Artists

摘要

It is found that most users have a favored score based on the dataset provided by Yahoo!, which is used most frequently in music ratings. We introduce the polarization ratio, indicating the frequency of a user's most favored score since the rating distributions depict apparent polarizations: 66.99% of the total users have the polarization ratio larger than 0.5; while 80.37% of the total users' most favored score is either 90 or 0. Our statistical analysis shows that the rating score distributions are highly polarized for average raters as well as the so called heavy raters regardless of the total times a user rated. Our analysis of the ratings on four categories of music items, tracks, albums, artists and genres, reveals the emergence of popular artists, which signifies the importance of the introduction of a user's most favored score and its polarization ratio in quantitative descriptions of human behavior in music ratings.