Please wait a minute...
Advanced Search
数据分析与知识发现  2021, Vol. 5 Issue (2): 94-105     https://doi.org/10.11925/infotech.2096-3467.2020.0521
  研究论文 本期目录 | 过刊浏览 | 高级检索 |
基于多源信息融合的音乐推荐方法 *
李丹阳,甘明鑫()
北京科技大学经济管理学院 北京 100083
Music Recommendation Method Based on Multi-Source Information Fusion
Li Danyang,Gan Mingxin()
School of Economics and Management, University of Science and Technology Beijing, Beijing 100083, China
全文: PDF (1639 KB)   HTML ( 19
输出: BibTeX | EndNote (RIS)      
摘要 

【目的】 利用多源信息融合构建音乐特征体系,解决音乐推荐冷启动问题,为用户提供个性化音乐推荐。【方法】 采用基于多源信息融合的两段式推荐模型。通过神经网络融合多源信息,构建音乐特征体系,预测音乐的潜在因子向量,从而解决音乐冷启动问题,实现TopN推荐。【结果】 在百万歌曲数据集上开展实验,所提出的方法与CNN模型相比,在F1值上的提升幅度达到9.13%,在RMSE、MAE上的降低幅度分别达到8.08%和3.91%。【局限】 两段式推荐方法较端到端的训练有更大的局限性;此外,使用梅尔频谱训练占用内存资源较高。【结论】 所提方法构建音乐特征体系,解决了音乐推荐冷启动问题,提高了音乐推荐性能。

服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
李丹阳
甘明鑫
关键词 音乐推荐深度学习信息融合神经网络    
Abstract

[Objective] This paper creates a musical feature system based on multi-source information, aiming to address the cold start issue facing music recommendation and provide personalized services. [Methods] We proposed a two-stage model with multi-source information fused by neural network algorithm. Then, we built the musical feature system and predicted the potential factor vectors of music. Finally, we generated the TopN recommendation list for the users. [Results] We examined our model with the Million Song Dataset. Compared with other models such as CNN, the F1 value was improved by 9.13%, and the RMSE, MAE values were reduced by 8.08% and 3.91%, respectively. [Limitations] Our new method encounters more limits than the end-to-end training ones. And training with the Mel-frequency spectrum demands much more memory. [Conclusions] The proposed model improves the performance of music recommendation services.

Key wordsMusic Recommendation    Deep Learning    Information Fusion    Neural Network
收稿日期: 2020-06-04      出版日期: 2021-03-11
ZTFLH:  G350  
基金资助:*国家自然科学基金项目(71871019);*国家自然科学基金项目(71471016);*国家自然科学基金项目(71531013)
通讯作者: 甘明鑫 ORCID:0000-0001-8751-0780     E-mail: ganmx@ustb.edu.cn
引用本文:   
李丹阳, 甘明鑫. 基于多源信息融合的音乐推荐方法 *[J]. 数据分析与知识发现, 2021, 5(2): 94-105.
Li Danyang, Gan Mingxin. Music Recommendation Method Based on Multi-Source Information Fusion. Data Analysis and Knowledge Discovery, 2021, 5(2): 94-105.
链接本文:  
https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/10.11925/infotech.2096-3467.2020.0521      或      https://manu44.magtech.com.cn/Jwk_infotech_wk3/CN/Y2021/V5/I2/94
Fig.1  音乐推荐方法总体框架
Fig.2  神经网络模型结构
Fig.3  音频数据处理流程
Fig.4  LSTM单元结构
Track_id Title Song_id Release
TRACCVZ128F4291A8A Deep Sea Creature SOVLGJY12A8C13FBED Call of the Mastodon
TRACCMH128F428E4CD No Quieras Marcharte SOGDQZK12A8C13F37C Adelante
TRACCJA128F149A144 Segredo SODPNJR12A6D4FA52D Joao Voz E Violato
TRACCUS128F92E1FEB Bedroom Acoustics SOMMSMW12A8C13FCCC Plug In Baby
TRACJDX12903CD4917 Church Hangover SOSJVGP12AB0185B95 El Mas Chingon
Artist_id Artist_mbid Artist_name Year
ARMQHX71187B9890D3 bc5e2ad6-0a4a-4d90-b911-e9a7e6861727 Mastodon 2001
AR2PT4M1187FB55B1A d54ea4a6-0b9c-4e47-bed0-289ae9ff4037 Los Chichos 1984
AR3THYK1187B999F1F 286ec4c2-b5ca-4f85-b331-280a6d73dd14 Jo?o Gilberto 2000
ARR3ONV1187B9A2F59 fd857293-5ab8-40de-b29e-55a69d4e4d0f Muse 0
ARYF20K1187B9B76BD 4a41c200-153b-4158-87b3-db0fea627e2c George Lopez 2006
Table 1  音乐元数据信息表
User_id Song_id Play_num
d3f65887d5eb4eac8ca14d0f3085c5021e79a8c5 SOAWFKO129F06933A7 2
4fd90e488869fffe6e2186a489311bdb19ebce46 SOUGUKH12A8C13FBB0 22
8e62f55bc5830dfcbe92339afd86fda1d7fccb5e SOOQUVZ12A6D4F7536 3
747155a884e6c96299a1a052e89ead19f31ad235 SONDICG12AB0184D30 6
3446fc0b811c4970eafe30866f873e44567d371c SOODZFJ12A6D4F8D03 1
Table 2  用户-音乐收听记录表
Artist_id Tag
AR002UA1187B9A637D garage rock
AR002UA1187B9A637D free jazz
AR002UA1187B9A637D pop
AR003FB1187B994355 rock
AR003FB1187B994355 hip hop
Table 3  歌手标签表
模型 MSE RMSE MAE MSLE
CNN 1.0 156 0.8 734 0.4 580 0.1 099
Year 1.0 129 0.8 727 0.4 599 0.1 098
CNN+LSTM 0.9 668 0.9 450 0.4 833 0.1 073
Category 0.9 648 0.8 580 0.4 548 0.1 035
Artist 0.8 821 0.8 232 0.4 457 0.0 921
本文方法 0.7 942 0.8 028 0.4 401 0.0 894
Table 4  误差对比
Fig.5  准确率结果
Fig.6  召回率结果
Fig.7  F1值结果
模型 准确率 召回率 F1
N=3 N=4 N=3 N=4 N=3 N=4
CNN+LSTM 0.6 075 0.5 543 0.7 160 0.8 229 0.5 992 0.6 065
CNN 0.6 104 0.5 552 0.7 237 0.8 307 0.6 038 0.6 094
Year 0.6 269 0.5 612 0.7 517 0.8 480 0.6 230 0.6 176
Category 0.6 445 0.5 739 0.7 746 0.8 661 0.6 426 0.6 318
Artist 0.6 595 0.5 867 0.7 724 0.8 665 0.6 491 0.6 397
本文方法 0.6 682 0.5 888 0.7 843 0.8 711 0.6 589 0.6 431
Table 5  推荐评价指标对比
Fig.8  音乐三维聚类图
Fig.9  潜在因子向量可视化结果
[1] Zheng E, Kondo G Y, Zilora S, et al. Tag-aware Dynamic Music Recommendation[J]. Expert Systems with Applications, 2018,106:244-251.
doi: 10.1016/j.eswa.2018.04.014
[2] Hu Y F, Koren Y, Volinsky C. Collaborative Filtering for Implicit Feedback Datasets[C]//Proceedings of the 8th IEEE International Conference on Data Mining. 2008: 263-272.
[3] van den Oord A, Dieleman S, Schrauwen B. Deep Content-based Music Recommendation[C]//Proceedings of the 26th International Conference on Neural Information Processing Systems. 2013: 2643-2651.
[4] Bertin-Mahieux T, Ellis P W D, Whitman B, et al. The Million Song Dataset[C]//Proceedings of the 12th International Society for Music Information Retrieval Conference. 2011: 591-596.
[5] 牛滨, 孔令志, 罗森林, 等. 基于MFCC和GMM的个性音乐推荐模型[J]. 北京理工大学学报, 2009,29(4):351-355.
[5] ( Niu Bin, Kong Lingzhi, Luo Senlin, et al. Individuality Music Recommendation Model Based on MFCC and GMM[J]. Transactions of Beijing Institute of Technology, 2009,29(4):351-355.)
[6] Liu C L, Chen Y C. Background Music Recommendation Based on Latent Factors and Moods[J]. Knowledge-Based Systems, 2018,159:158-170.
doi: 10.1016/j.knosys.2018.07.001
[7] Li T, Choi M, Fu K M, et al. Music Sequence Prediction with Mixture Hidden Markov Models[C]//Proceedings of 2019 IEEE International Conference on Big Data. Los Angeles, CA, USA, 2018. DOI: 10.1109/BigData47090.2019.9005695.
[8] Flexer A, Stevens J. Mutual Proximity Graphs for Improved Reachability in Music Recommendation[J]. Journal of New Music Research, 2018,47(1):17-28.
doi: 10.1080/09298215.2017.1354891 pmid: 29348779
[9] McFee B, Barrington L, Lanckriet G. Learning Content Similarity for Music Recommendation[J]. IEEE Transactions on Audio, Speech, and Language Processing, 2012,20(8):2207-2218.
doi: 10.1109/TASL.2012.2199109
[10] Lecun Y, Bottou L, Bengio Y, et al. Gradient-based Learning Applied to Document Recognition[J]. Proceedings of the IEEE, 1998,86(11):2278-2324.
doi: 10.1109/5.726791
[11] Krizhevsky A, Sutskever I, Hinton G E. ImageNet Classification with Deep Convolutional Neural Networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. 2012: 1097-1105.
[12] Simonyan K, Zisserman A. Very Deep Convolutional Networks for Large-Scale Image Recognition[C]//Proceedings of International Conference on Learning Representations, 2015.
[13] Lee J, Lee K, Park J, et al. Deep Content-User Embedding Model for Music Recommendation[OL]. arXiv Preprint, arXiv: 1807. 06786.
[14] Hochreiter S, Schmidhuber J. Long Short-Term Memory[J]. Neural Computation, 1997,9(8):1735-1780.
doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[15] Chung J, Gulcehre C, Cho K, et al. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling [OL]. arXiv Preprint, arXiv: 1412. 3555.
[16] Balakrishnan A, Dixit K. DeepPlaylist: Using Recurrent Neural Networks to Predict Song Similarity [EB/OL]. [2019-05-16].https://cs224d.stanford.edu/reports/BalakrishnanDixit.pdf.
[17] Sachdeva N, Gupta K, Pudi V. Attentive Neural Architecture Incorporating Song Features for Music Recommendation[C]//Proceedings of the 12th ACM Conference on Recommender Systems. 2018: 417-421.
[18] Wang D J, Deng S G, Xu G D. Sequence-Based Context-Aware Music Recommendation[J]. Information Retrieval, 2018,21(2/3):230-252.
doi: 10.1007/s10791-017-9317-7
[19] Zou W. Design and Application of Incremental Music Recommendation System Based on Slope One Algorithm[J]. Wireless Personal Communications, 2018,102(4):2785-2795.
doi: 10.1007/s11277-018-5303-7
[20] Deng S G, Wang D J, Li X T, et al. Exploring User Emotion in Microblogs for Music Recommendation[J]. Expert Systems with Applications, 2015,42(23):9284-9293.
doi: 10.1016/j.eswa.2015.08.029
[21] Ren J, Kauffman R, King D. Two-Sided Value-Based Music Artist Recommendation in Streaming Music Services[C]//Proceedings of the 52nd Hawaii International Conference on System Sciences. 2019: 2679-2688.
[22] Chen J P, Ying P G, Zou M. Improving Music Recommendation by Incorporating Social Influence[J]. Multimedia Tools and Applications, 2019,78(3):2667-2687.
doi: 10.1007/s11042-018-5745-7
[23] Cheng Z Y, Shen J L. On Effective Location-Aware Music Recommendation[J]. ACM Transactions on Information Systems, 2016, 34(2): Article 13.
[24] 李洋, 赵鸣, 徐梦瑶, 等. 多源信息融合技术研究综述[J]. 智能计算机与应用, 2019,9(5):186-189.
[24] ( Li Yang, Zhao Ming, Xu Mengyao, et al. A Survey of Research on Multi-Source Information Fusion Technology[J]. Intelligent Computer and Applications, 2019,9(5):186-189.)
[25] Mikolov T, Chen K, Corrado G, et al. Efficient Estimation of Word Representations in Vector Space[C]//Proceedings of Workshop at International Conference on Learning Representations, 2013.
[1] 范少萍,赵雨宣,安新颖,吴清强. 基于卷积神经网络的医学实体关系分类模型研究*[J]. 数据分析与知识发现, 2021, 5(9): 75-84.
[2] 周泽聿,王昊,赵梓博,李跃艳,张小琴. 融合关联信息的GCN文本分类模型构建及其应用研究*[J]. 数据分析与知识发现, 2021, 5(9): 31-41.
[3] 范涛,王昊,吴鹏. 基于图卷积神经网络和依存句法分析的网民负面情感分析研究*[J]. 数据分析与知识发现, 2021, 5(9): 97-106.
[4] 顾耀文, 张博文, 郑思, 杨丰春, 李姣. 基于图注意力网络的药物ADMET分类预测模型构建方法*[J]. 数据分析与知识发现, 2021, 5(8): 76-85.
[5] 张乐, 冷基栋, 吕学强, 崔卓, 王磊, 游新冬. RLCPAR:一种基于强化学习的中文专利摘要改写模型*[J]. 数据分析与知识发现, 2021, 5(7): 59-69.
[6] 赵丹宁,牟冬梅,白森. 基于深度学习的科技文献摘要结构要素自动抽取方法研究*[J]. 数据分析与知识发现, 2021, 5(7): 70-80.
[7] 徐月梅, 王子厚, 吴子歆. 一种基于CNN-BiLSTM多特征融合的股票走势预测模型*[J]. 数据分析与知识发现, 2021, 5(7): 126-138.
[8] 钟佳娃,刘巍,王思丽,杨恒. 文本情感分析方法及应用综述*[J]. 数据分析与知识发现, 2021, 5(6): 1-13.
[9] 黄名选,蒋曹清,卢守东. 基于词嵌入与扩展词交集的查询扩展*[J]. 数据分析与知识发现, 2021, 5(6): 115-125.
[10] 马莹雪,甘明鑫,肖克峻. 融合标签和内容信息的矩阵分解推荐方法*[J]. 数据分析与知识发现, 2021, 5(5): 71-82.
[11] 韩普,张展鹏,张明淘,顾亮. 基于多特征融合的中文疾病名称归一化研究*[J]. 数据分析与知识发现, 2021, 5(5): 83-94.
[12] 孟镇,王昊,虞为,邓三鸿,张宝隆. 基于特征融合的声乐分类研究*[J]. 数据分析与知识发现, 2021, 5(5): 59-70.
[13] 张国标,李洁. 融合多模态内容语义一致性的社交媒体虚假新闻检测*[J]. 数据分析与知识发现, 2021, 5(5): 21-29.
[14] 王楠,李海荣,谭舒孺. 基于改进SMOTE算法与集成学习的舆情反转预测研究*[J]. 数据分析与知识发现, 2021, 5(4): 37-48.
[15] 常城扬,王晓东,张胜磊. 基于深度学习方法对特定群体推特的动态政治情感极性分析*[J]. 数据分析与知识发现, 2021, 5(3): 121-131.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
版权所有 © 2015 《数据分析与知识发现》编辑部
地址:北京市海淀区中关村北四环西路33号 邮编:100190
电话/传真:(010)82626611-6626,82624938
E-mail:jishu@mail.las.ac.cn