由於音樂播放平台的多樣性，使得人們能夠隨時隨地聆聽各種音樂。然而普通的音樂播放軟體(例如Microsoft Media Player 和Winamp Media Player)仍然需要使用者主動組織曲目與播放清單。隨著曲目數量的快速上升，人們將會需要具備有效性與智能性音樂推薦系統來幫助他們找到想聽的音樂。另外，使用者對於聆聽音樂的興趣常常會依照他們所處在的情境與進行中的活動而有所差別。我們在音樂推薦的方向中提出一個架構，採用了”在協同式過濾中基於情境的物件評分分離”與”機率性矩陣因子化”方法。為了收集帶有真實情境資訊的使用者音樂播放記錄，我們也實做了一個線上音樂播放平台，偵測並收集了多種真實的使用者情境資訊。實驗結果顯示在我們所提出的架構之下，預測的效果會比只使用目前最先進的方法其中之一─機率性矩陣因子化─更好並且在統計上具有顯著的差異。本實驗結果顯示情境化的資料有助於改善協同式過濾系統應用在音樂推薦上的效能。

There are various music player platforms enabling people to listen to music anywhere and anytime. However, common player software (e.g., Microsoft Media Player and Winamp Media Player) still need user to organize songs and playlist. As the amount of music increases, people will need music recommendation systems (MRS) that retrieve music intelligently and efficiently to help them find songs they like to listen. In addition, users usually alter their preferences of music under different contexts/activities. We propose a music recommendation approach that exploits “context-based splitting of item ratings in collaborative filtering” and “probabilistic matrix factorization (PMF)”. To collect the music listening record with real contextual information, we implement an online music player platform. The platform could detect and record real contextual information of users. The experimental results show that our approach outperforms PMF-based system, which is one of the state-of-the-art methods, with a statistically significant difference. The results demonstrate that context computing should be employed in music recommendation systems. We also analyze which types of contexts are the key factors of system performance.

書名頁 i
論文口試委員審定書 ii
授權書 iii
中文摘要 iv
英文摘要 v
誌謝 vii
目錄 viii
表目錄 x
圖目錄 x
Chapter 1 Introduction 1
Chapter 2 Related Work 3
Chapter 3 Notations and Preliminaries 5
3.1　Notations 5
3.2　Probabilistic Matrix Factorization 5
3.3　Item splitting 6
Chapter 4 Methodology 8
4.1　Architecture 8
Chapter 5 Experimental Evaluation 10
5.1　Platform and Dataset 10
5.2　Evaluation Metric 13
5.3　Impact of Contextual Information 13
Chapter 6 Conclusion and Future Work 17
Reference 19

[1]http://agent.cse.tw/youtube.
[2]G. Adomavicius and A. Tuzhilin, "Context-aware recommender systems," presented at the Proceedings of the 2008 ACM conference on Recommender systems, Lausanne, Switzerland, 2008.
[3]L. Baltrunas and F. Ricci, "Context-based splitting of item ratings in collaborative filtering," in Proceedings of the third ACM conference on Recommender systems, New York, New York, USA, 2009, pp. 245-248.
[4]R. M. Bell and Y. Koren, "Scalable collaborative filtering with jointly derived neighborhood interpolation weights," 2007, pp. 43-52.
[5]A. K. Dey, "Understanding and Using Context," Personal Ubiquitous Comput., vol. 5, pp. 4-7, 2001.
[6]C.-C. Hung, et al., "Tag-based User Profiling for Social Media Recommendation," in Workshop on Intelligent Techniques for Web Personalization and Recommender Systems, AAAI 2008, 2008.
[7]Y. Koren, "Factorization meets the neighborhood: a multifaceted collaborative filtering model," presented at the Proceeding of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, Las Vegas, Nevada, USA, 2008.
[8]Y. Koren, et al., "Matrix Factorization Techniques for Recommender Systems," Computer, vol. 42, pp. 30-37, 2009.
[9]U. Panniello, et al., "Experimental comparison of pre- vs. post-filtering approaches in context-aware recommender systems," in Proceedings of the third ACM conference on Recommender systems, New York, New York, USA, 2009, pp. 265-268.
[10]M. Pazzani and D. Billsus, "Content-based recommendation systems," The adaptive web, pp. 325-341, 2007.
[11]R. Salakhutdinov and A. Mnih, "Probabilistic matrix factorization," Advances in neural information processing systems, vol. 20, p. 1257¡V1264, 2008.
[12]T. Strang and C. Linnhoff-Popien, "A context modeling survey," 2004.
[13]Y. Zhen, et al., "TagiCoFi: tag informed collaborative filtering," in Proceedings of the third ACM conference on Recommender systems, New York, New York, USA, 2009, pp. 69-76.
[14]T. C. Zhou, et al., "UserRec: A User Recommendation Framework in Social Tagging Systems," 2010.