Responsive image
博碩士論文 etd-0812113-095859 詳細資訊
Title page for etd-0812113-095859
論文名稱
Title
結合音樂內容和協同過濾方法之音樂推薦
Combining Content-base and Collaborative Filtering Methods for Music Recommendation
系所名稱
Department
畢業學年期
Year, semester
語文別
Language
學位類別
Degree
頁數
Number of pages
67
研究生
Author
指導教授
Advisor
召集委員
Convenor
口試委員
Advisory Committee
口試日期
Date of Exam
2014-06-26
繳交日期
Date of Submission
2014-09-09
關鍵字
Keywords
音樂資訊擷取、矩陣分解、音樂標籤、推薦系統、協同式推薦
Music Information Retrieval, Matrix Decomposition, Recommendation System, Collaborative Filtering, Music Tags
統計
Statistics
本論文已被瀏覽 5765 次,被下載 907
The thesis/dissertation has been browsed 5765 times, has been downloaded 907 times.
中文摘要
伴隨音樂訊號處理的進步以及網路的演進,音樂除了實體唱片與廣播等傳播方式之外,更有網路串流和行動裝置等多元化之聆聽方式,在音樂的推薦上除了透過歌曲的音訊分析外,亦有利用音樂的類型和使用者撥放紀錄等額外之資訊進行演算,在現行音樂庫不斷快速增加的情況下,協助使用者找到可能喜歡的歌曲也是現行的主流音樂服務之一。
在網路串流音樂環境中,本研究將聆聽紀錄視為使用者的偏好習慣,並將之轉換為使用者對該音樂的評分,將之作為推薦基礎,進行演算,並依照現有的音樂資訊擷取領域之推薦系統的幾個要素:音訊、音樂標籤、協同式方法,分別實作音樂內容、協同式處理方法,在音樂內容上嘗試將音訊和音樂標籤進行篩選以及分類,並提出結合分類屬性的改良方法;在協同式推薦方面則採用最近鄰居方法和矩陣分解,並帶入門閥值和權重值進行調整,透過這兩種不同類型的推薦系統方法,找出使用者最感興趣之音樂項目。對照過去研究所提出的推薦方法,上述兩種之音樂推薦之改良方法皆得出較高的精確度。
Abstract
With the music signal processing and network development, people can listen to music not only CDs, radio but also mobile devices, even more, online streaming.
Music recommendation systems use audio signals as classification features to predict user prefferences.In the recent years, music metadata such as music genre, music tags, and playlists can be treated as attribute to build recommendation models. The music database is growing explosively so that it is hard for users to find exactly what they want. Nowadays, there are many music services have recommendation systems for their users to find out this potential ‘songs’ quickly.
In online streaming environment, this research considers user playlists as user preferences, and converts playcounts to user preference rates. By this step, rates play a role in the classification and collaborative filtering methods. In music information retrieval field, content-based(audio, music tags) and collaborative approach are well known and frequently implemented in music services. In this research, content-based approach uses music signals and tag as classification features, and tries to combine both attributes into classification model; on the other hand, collaborative approach, uses k-nearest neighborhood method and matrix decomposition by tuning thresholds and vector weights. Both of these approachs are focusing on music items which user inetersted. According to the experement, approached methods accuracies are higher than traditional ones.
目次 Table of Contents
論文審定書 i
摘要 ii
Abstract iii
第一章 緒論 1
1.1 研究背景 1
1.2 研究動機與目的 2
1.3 研究流程 3
第二章 文獻探討 4
2.1 音樂內容相似度 4
2.1.1 音樂特徵擷取 4
2.1.2 相似度量測 5
2.1.3 音樂標籤 5
2.2 協同過濾式推薦 6
2.2.1 KNN 6
2.2.2 SVD 7
第三章 研究方法與步驟 9
3.1 系統架構 9
3.1.1 Last.fm Dataset 9
3.1.2 資料前處理 10
3.1.3 基於內容的方法 10
3.1.4 協同過濾式方法 10
3.1.5 推薦系統 10
3.2 資料集介紹與處理 11
3.2.1 Last.fm Dataset 11
3.2.2 Million Song Dataset 12
3.2.3 資料處理 13
3.3 音樂內容方法實作 15
3.3.1 音樂訊號 15
3.3.2 音樂標籤 19
3.3.2.1 標記數量 19
3.3.2.2 音樂風格 20
3.3.2.3 個人情緒 21
3.3.2.4 歌手類型 21
3.4 協同過濾方法實作 22
3.4.1 KNN系統實作 22
3.4.2 SVD系統實作 25
3.5 結合音樂音訊和音樂標籤之分類 27
3.5.1 音樂訊號結合音樂標籤(數量型) 28
3.5.2 音樂訊號結合音樂標籤(歌手型) 30
3.6 改良協同過濾法 32
3.6.1 KNN-相似度過濾法 32
3.6.2 SVD-相似度權重法 33
3.6.3 SVD-標籤調整法 34
3.6.4 SVD-標籤權重混合法 35
第四章 實驗與結果 36
4.1 方法評量與績效 36
4.1.1 音樂訊號結合音樂標籤(數量型) 36
4.1.2 音樂訊號結合音樂標籤(歌手型) 38
4.1.3 KNN-相似度過濾法 39
4.1.4 SVD-相似度權重法 41
4.1.5 SVD-標籤調整法 42
4.1.6 SVD-標籤權重混合法 43
4.2 研究比較與分析 44
4.3 情境模擬 51
第五章 結論 54
5.1 研究貢獻 54
5.2 未來展望 55
第六章 參考文獻 56
參考文獻 References
[1] Andrew,etal. "Methods and metrics for cold-start recommendations." Proceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval. ACM, 2002.
[2] Alicja, Piotr Synak, and Zbigniew W. Raś. "Multi-label classification of emotions in music." Intelligent Information Processing and Web Mining. Springer Berlin Heidelberg, 2006. 307-315.
[3] Cohen, William W., and Wei Fan. "Web-collaborative filtering: Recommending music by crawling the web." Computer Networks 33.1 (2000): 685-698.
[4] Douglas, Eck, et al. "Automatic generation of social tags for music recommendation." Advances in neural information processing systems. 2008.
[5] Domingues, Marcos Aurélio, et al. "Combining usage and content in an online recommendation system for music in the long tail." International Journal of Multimedia Information Retrieval 2.1 (2013): 3-13.
[6] Evangelos, Moutselakis V., and Alexandros S. Karakos. "Semantic Web Multimedia Metadata Retrieval: A Music Approach." Informatics, 2009. PCI'09. 13th Panhellenic Conference on. IEEE, 2009.
[7] Eck, Douglas, et al. "Automatic generation of social tags for music recommendation." Advances in neural information processing systems. 2008.
[8] Fazilatur, Rahman and Jawed Siddiqi. "Semantic annotation of digital music."Journal of Computer and System Sciences 78.4 (2012): 1219-1231.
[9] George, Tzanetakis, and Perry Cook. "Musical genre classification of audio signals." Speech and Audio Processing, IEEE transactions on 10.5 (2002): 293-302.
[10] Han-Saem, Park, Ji-Oh Yoo, and Sung-Bae Cho. "A context-aware music recommendation system using fuzzy bayesian networks with utility theory."Fuzzy systems and knowledge discovery. Springer Berlin Heidelberg, 2006. 970-979.
[11] Hinton, Geoffrey, et al. "Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups." Signal Processing Magazine, IEEE 29.6 (2012): 82-97.
[12] http://last.fm/
[13] http://labrosa.ee.columbia.edu/millionsong/lastfm
[14] http://www.music-ir.org/mirex/wiki/2008:Audio_Tag_Classification
[15] https://www.spotify.com/
[16] Huang, Yu-siang. "Design and Implementation for Content-based Singer Classification on Compressed Domain Audio Data." (2010).
[17] Jiang, Dan-Ning, et al. "Music type classification by spectral contrast feature."Multimedia and Expo, 2002. ICME'02. Proceedings. 2002 IEEE International Conference on. Vol. 1. IEEE, 2002.
[18] Johnson, Julene K., et al. "Music recognition in frontotemporal lobar degeneration and Alzheimer disease." Cognitive and behavioral neurology: official journal of the Society for Behavioral and Cognitive Neurology 24.2 (2011): 74.
[19] Konstantinos, Trohidis, et al. "Multi-Label Classification of Music into Emotions." ISMIR. Vol. 8. 2008.
[20] Koren, Yehuda, Robert Bell, and Chris Volinsky. "Matrix factorization techniques for recommender systems." Computer 42.8 (2009): 30-37.
[21] Koren, Yehuda. "Factorization meets the neighborhood: a multifaceted collaborative filtering model." Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2008.
[22] Koren,Yehuda,andRobert Bell. "Advances in collaborative filtering."Recommender Systems Handbook. Springer US, 2011. 145-186.
[23] Lee, Chi-Chun, et al. "Emotion recognition using a hierarchical binary decision tree approach." Speech Communication 53.9 (2011): 1162-1171.
[24] Lew, Michael S., et al. "Content-based multimedia information retrieval: State of the art and challenges." ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP) 2.1 (2006): 1-19.
[25] Matthew D., Hoffman, David M. Blei, and Perry R. Cook. "Easy As CBA: A Simple Probabilistic Model for Tagging Music." ISMIR. Vol. 9. 2009.
[26] Mandel, Michael I., Graham E. Poliner, and Daniel PW Ellis. "Support vector machine active learning for music retrieval." Multimedia systems 12.1 (2006): 3-13.
[27] Nanopoulos, Alexandros, et al. "Musicbox: Personalized music recommendation based on cubic analysis of social tags." Audio, Speech, and Language Processing, IEEE Transactions on 18.2 (2010): 407-412.
[28] Ness, Steven R., et al. "Improving automatic music tag annotation using stacked generalization of probabilistic svm outputs." Proceedings of the 17th ACM international conference on Multimedia. ACM, 2009.
[29] Nilsson, Martin. "ID3 tag version 2.3. 0." Informal Standard id3v2 3 (1999).
[30] Piccart, Beau, Jan Struyf, and Hendrik Blockeel. "Alleviating the Sparsity Problem in Collaborative Filtering by Using an Adapted Distance and a Graph-Based Method." SDM. 2010.
[31] Sanjunath, Bangalore S., Philippe Salembier, and Thomas Sikora, eds.Introduction to MPEG-7: multimedia content description interface. Vol. 1. John Wiley & Sons, 2002.
[32] Song, Seheon, et al. "Music ontology for mood and situation reasoning to support music retrieval and recommendation." Digital Society, 2009. ICDS'09. Third International Conference on. IEEE, 2009.
[33] Su, Xiaoyuan, and Taghi M. Khoshgoftaar. "A survey of collaborative filtering techniques." Advances in artificial intelligence 2009 (2009): 4.
[34] Tatli, Ipek, and Aysenur Birturk. "A tag-based hybrid music recommendation system using semantic relations and multi-domain information." Data Mining Workshops (ICDMW), 2011 IEEE 11th International Conference on. IEEE, 2011.
[35] Teng, Yuan-Ching, Ying-Shu Kuo, and Yi-Hsuan Yang. "A large in-situ dataset for context-aware music recommendation on smartphones." Multimedia and Expo Workshops (ICMEW), 2013 IEEE International Conference on. IEEE, 2013.
[36] Turnbull, Douglas, et al. "Semantic annotation and retrieval of music and sound effects." Audio, Speech, and Language Processing, IEEE Transactions on 16.2 (2008): 467-476.
[37] Yves, Raimond, and Samer Abdallah. The event ontology. Technical report, 2007. http://motools. sourceforge. net/event, 2007.
[38] Zhang, Yibin, and Jie Zhou. "A study on content-based music classification."Signal Processing and Its Applications, 2003. Proceedings. Seventh International Symposium on. Vol. 2. IEEE, 2003.
電子全文 Fulltext
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available


紙本論文 Printed copies
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。
開放時間 available 已公開 available

QR Code