基于艺人画像的歌曲点播量预测

摘要

用户画像可以用于用户相关事件的预测.在本文中,我们探索对用户画像的建模中结合外在因素的变化,对用户相关行为进行预测.在实验方面,我们以阿里音乐用户的历史播放数据为基础,结合外在主要相关事件,对艺人进行多维的画像,进而预测某个时间段内艺人的歌曲点播量,挖掘出即将成为潮流的艺人,从而实现对一个时间段内音乐流行趋势的准确把控.实验结果表明,与传统的机器学习方法和时序模型方法相比,我们的算法具有很高准确率,且具有简洁,泛化能力强的特点.%User profiling can be used to predict user related events.In this paper,we explore the combination of user profiling and environmental factors to improve the accuracy of user related event prediction.In experiments,we use Ali music data to test our hypothesis.While profiling the users in the dataset,we consider both related external events and users' own features into categorization.We thereafter predict the popularity of the artists' artworks based on these profiling.Experimental results show that compared with the traditional machine learning methods and time series models,our algorithm has high accuracy.It also shows that the proposed method has the advantage of simplicity and good generalization.