Research on the Influence Factors and Genre Development Trends of Music Based on PageRank and LSTM Model

Abstract

With the development of the new era, music is also developing. When artists create new music, many factors affect them. How to quantify the impact factor and predict the development trend of different genres during the evolution of music is a significant research topic in the current society. Based on the relationship between influencers and followers in the influence_data dataset provided by Integrative Collective Music (ICM), this paper establishes a directed network graph structure to express music influence and uses the PageRank algorithm to dynamically solve the influence degree of influencers on followers to quantify. Impact factor development captures the parameters of music influence in this network. And according to the aggregated influence_data dataset, the LSTM (Long Short-term Memory) neural network is established to predict different music genres development trend. The three types of MAE predicted (Pop/Rock, Country, Jazz), MSE and R2 are MAE (6.675) ,7.843,8.306),MSE(71.879,108.297,103.521),R2(0.762,0.659 ,0.676). The results show that the model can quantify the impact factor very well and achieve good results under these three evaluation indicators.