User Modeling in Music Interest Based on Wavelet Support Vector Machine

摘要：

This paper proposes an approach of user model-ing in music interest based on wavelet kernel support vector machine algorithm. The approach, making use of the characteristics of wavelet kernel which can acquire the low frequency signal and deal with the high frequency signal, can construct the more accu-rate separating hyper-plane for user interest, and thus create more effective the model of personalized mu-sic interest. The experiment shows that the model created by the approach has the better prediction performance than that of the model based on deci-sion tree, BP and radial basic function kernel support vector machine. In addition, the approach is capable of rapid prototyping in condition of small training sets and can solve the problem of difficulties in user mod-eling caused by sparse samples which occurred in the first operating period of personalized service system.

作者: Ruoyu Fang Weiling Xu Qiong Zhang Hong Zhang Xing Lin Haoxun Li

作者单位: Guangdong Provincial Key Laboratory of Digital Signal and Image Processing Technology,Shantou Univer Information College,Zhongkai University of Agriculture and Engineering,Guangzhou 510225, China Guangdong Provincial Key Laboratory of Digital Signal and Image Processing Technology,Shantou Univer College of Science,Shantou University,Shantou 515063, China Department of Computer Science,Shantou University,Shantou 515063, China Department of Computer Science, Shantou University, Shantou 515063, China

会议类型: 国际会议

会议名称: 2010 China-Ireland International Conference on Information and Communications Technologies(2010年中国-爱尔兰信息与通信技术国际会议 CIICT 2010)

会议地点: 武汉

会议语种:英文

页码: 93-98

在线出版日期: 2010-10-10（万方平台首次上网日期，不代表论文的发表时间）