融合音乐感情注意力的长短期音乐偏好推荐算法研究  

Research on the Application of High-quality Smart Classroom Education Evaluation in theInternet of Things Environment

作　　者：王珂欣 Wang Kexin(Shaanxi Vocational Academy Of Art,Conservatory of music,Shaanxi Xi’an 710061)

机构地区：[1]陕西艺术职业学院音乐学院,陕西西安710061

出　　处：《现代科学仪器》2024年第5期287-292,共6页Modern Scientific Instruments

摘　　要：随着居民娱乐方式的升级,基于感情注意力的音乐推荐算法得以发展。传统的音乐推荐算法基于用户的历史听歌记录,但忽视了音乐本身对用户情感的影响。因此此次研究将梅尔滤波器组入用户长短期偏好中,并在考虑感情注意力的基础上对序列时段进行分析,生成一种音乐推荐算法。研究的实验在Recom数据集上进行,并同时进行随机森林等三种算法的实验,以验证融合算法的有效性。针对多曲目的判别能力实验,融合算法对于用户心理的判断准确率为97%,在四种算法中表现最好。实验结果表明,研究提出的融合算法具有最强的性能,适于在用户偏好音乐推荐中得到应用。With the upgrading of residents'entertainment methods,music recommendation algorithms based on emotional attention have been developed.The traditional music recommendation algorithm is based on the user's historical listening records,but ignores the impact of the music itself on the user's emotions.Therefore,in this study,Mel filters were incorporated into user long-term and short-term preferences,and a music recommendation algorithm was generated by analyzing the sequence time period while considering emotional attention.The research experiment was conducted on the Recom dataset,and experiments were conducted on three algorithms including random forest to verify the effectiveness of the fusion algorithm.For the experiment of discriminative ability for multiple tracks,the fusion algorithm has an accuracy rate of 97%in judging user psychology,and performs the best among the four algorithms.The experimental results show that the fusion algorithm proposed in the study has the strongest performance and is suitable for application in user preference music recommendation.

关 键 词：长短期偏好 音乐推荐 序列时段 梅尔滤波器组 感情注意力 

分 类 号：G613.5[文化科学—学前教育学]