Unveiling music genre structure through common-interest communities

Using a dataset of more than 90,000 metal music reviews written by over 9000 users in period 15 years, we analyse the genre structure with aid review text information. We model relationships between genres using user-oriented network, based on reviews. then perform community detection and employ network “averaging” method to obtain stable clusters, order structures clusters both locally within each cluster globally entire network. In addition identifying use Dependency Parsing modified Term Frequency–Inverse Document Frequency extract significant unique features cluster. These information can allow us understand how audience (fans) perceive similar different genres, also assist classifying which share common-interest user communities, offering objective way grouping genres. Furthermore, classification help recommendation engines provide targeted suggestions music, potentially musicians select labels for their design better cater preferences audiences previous