進階搜尋


   電子論文尚未授權公開,紙本請查館藏目錄
(※如查詢不到或館藏狀況顯示「閉架不公開」,表示該本論文不在書庫,無法取用。)
系統識別號 U0026-0208201816343300
論文名稱(中文) 基於使用者喜好建立時序性個人化推薦系統
論文名稱(英文) Modeling User Preference in Temporal Personalized Recommendation
校院名稱 成功大學
系所名稱(中) 資訊工程學系
系所名稱(英) Institute of Computer Science and Information Engineering
學年度 106
學期 2
出版年 107
研究生(中文) 劉曉筠
研究生(英文) Hsiao-Yun Liu
學號 P76051030
學位類別 碩士
語文別 英文
論文頁數 41頁
口試委員 指導教授-蔣榮先
口試委員-陳信希
口試委員-孫永年
口試委員-郝沛毅
口試委員-鄞宗賢
中文關鍵字 推薦系統  個人化推薦系統  長期喜好  短期喜好  深度學習  遞歸神經網路  卷積神經網絡 
英文關鍵字 Recommendation System  Personalized Recommendation System  Long-term Preference  Short-term Preference  Deep Learning  Recurrent Neural Networks  Convolutional Neural Networks 
學科別分類
中文摘要 在這個資訊量爆炸的時代,使用者對於資訊有更多樣性的選擇,進而促使推薦系統在近年來的蓬勃發展。然而,隨著使用者的需求及喜好快速變遷,要如何掌握使用者在不同時間的需求及喜好在推薦系統中是很重要的課題,因此,近年來不少研究致力於發展個人化的推薦系統來探索使用者的喜好,而在個人化推薦系統中考量時間因素也是值得研究的議題。
使用者喜好的探索在個人化推薦系統中扮演很重要的角色,以往對於使用者喜好的探索的技術在於利用使用者的個人檔案或靜態的歷史紀錄進行喜好的分析,但上述的技術無法針對得知使用者隨著時間變化的喜好,因此,本研究提出了一個新的方法來探索使用者時序性的喜好並應用於個人化推薦系統中。在本研究中對於使用者的喜好分成長期及短期兩種層面進行探索,因應長期喜好有著較穩定且不易改變的特性;且短期喜好則易受到時間及環境造成改變。因此,在本研究中針對使用者的長期及短期喜好給予不同的分析方式,更精準地捕捉使用者時序性的喜好。
本研究提出了一個基於深度學習方法的框架來實現時序性的個人化推薦,在框架中,本研究採用了RNN模型將使用者序列性的歷史紀錄進行短期喜好的編碼,以及利用CNN模型將使用者的序列性的歷史紀錄進行長期喜好的編碼,並將使用者的短期及長期喜好用三種不同的方式來合併以取得使用者時序性喜好的特徵,最後,將使用者喜好特徵進行解碼得到推薦結果。
在實驗中,本研究採用了Pinterest的資料集進行推薦結果的評估,實驗結果說明了本研究提出的方法不僅比傳統的方法有更好的推薦表現,並證明了同時考慮使用者長期及短期喜好的好處,以及本研究提出的兩階段訓練方式能提升推薦效果,而本研究在實驗的最後還提出了幾個有趣的個案研究來說明本研究的可用性及有效性。最後,期望本研究能應用於現實生活中,給予人們更適合的個人化推薦。
英文摘要 In the era of information explosion, the users have numerous choices about information. Recommendation systems have become increasingly popular in recent years. However, user’s preferences are changing all the time. How to capture user’s preferences at different times is a big issue in recommendation system. Therefore, there are many studies are dedicated to developing personalized recommendation systems to explore user’s preferences. Temporal factors for the users in personalized recommendation systems is also a necessary task.
It is important to explore user’s preferences in personalized recommendation systems. Many previous studies utilized user’s profiles or static historical records to know his/her preferences. However, the above works cannot capture the user’s interests changing. Therefore, our study proposes a novel personalized recommendation system to explore user’s temporal preferences. In our works, we model user’s preference for both short-term and long-term aspects. Due to the static property of long-term preference and the changeable property of short-term preference, we apply different methods to analyze user’s long-term and short-term behaviors for the task of capturing user’s preferences precisely.
Deep learning has become increasingly popular in recent years. Accordingly, our study proposes a deep learning based framework for temporal personalized recommendation. In the framework, we use RNN model to encode the user’s sequential behaviors to represent user’s short-term preference, while we apply CNN model to encode the user’s sequential behaviors to represent user’s long-term preference. By integrating the user’s short-term and long-term preferences, we adopt three combination mechanisms to get the user’s temporal feature. At last, we employ a decoder to convert the user’s temporal feature into a recommending score.
We conduct our experiments on the Pinterest dataset and evaluate the recommendation performance. The experimental results demonstrate that our proposed model outperforms the traditional methods and show the benefit of considering both user’s long-term and short-term preferences. Moreover, our proposed two-step training mechanism can enhance the recommendation performance. We provide several case studies to show the availability and effectiveness of our work. Finally, we hope that our personalized temporal recommendation system can be applied to real-world recommendation scenario.
論文目次 中文摘要 I
Abstract III
誌謝 V
Contents VI
List of Tables VIII
List of Figures IX
Chapter 1 Introduction 1
1.1 Background and Motivation 1
1.2 Research Objectives 2
1.3 Thesis Organization 3
Chapter 2 Related Work 4
2.1 Traditional methods of Recommendation System 4
2.1.1 Content-based Recommendation 4
2.1.2 Collaborative Filtering 5
2.2 Deep Learning based Recommendation System 5
2.2.1 Neural Network Recommendation 5
2.2.2 Recurrent Neural Networks 6
2.3 Temporal dynamics in Recommendation 6
Chapter 3 User Preference Modeling 8
3.1 Overview 8
3.2 Short-term Preference Encoder 10
3.3 Long-term Preference Encoder 11
3.4 Temporal Feature Generator 14
Chapter 4 Temporal Personalized Recommendation 16
4.1 Similarity Function 16
4.2 Model Training 17
4.3 Ranking and Recommendation 18
Chapter 5 Experimental Design and Results 19
5.1 Experimental Design 19
5.1.1 Dataset Description 19
5.1.2 Data Preprocessing 20
5.1.2.1 Filtering 21
5.1.2.2 Embedding 21
5.1.2.3 Data Enrichment 22
5.2 Baseline Methods 22
5.2.1 Matrix Factorization (MF) 22
5.2.2 Factorizing Personalized Markov Chains (FPMC) 23
5.3 Evaluation Metrics 23
5.3.1 Mean Reciprocal Rank (MRR) 24
5.3.2 Precision at k and Recall at k 24
5.4 Experimental Settings 25
5.5 Experimental Results and Analysis 28
5.5.1 Comparison among Different Combination Methods 29
5.5.2 Overall Performance 30
5.5.3 Influence of Two-step Training 31
5.6 Case Study 33
Chapter 6 Conclusion and Future Work 36
6.1 Conclusion 36
6.2 Future Work 37
Reference 39
參考文獻 [1]B.Hidasi, A.Karatzoglou, L.Baltrunas, and D.Tikk, “Session-based Recommendations with Recurrent Neural Networks,” Proceedings of the 4th. International Conference on Learning Representations, p. 425–434, 2016.
[2]M. J.Pazzani and D.Billsus, “Content-Based Recommendation Systems,” The Adaptive Web. Lecture Notes in Computer Science, vol. 4321, p. 325–341, 2007.
[3]G.Linden, B.Smith, and J.York, “Amazon.com recommendations: Item-to-item collaborative filtering,” IEEE Internet Computing, vol. 7, no. 1, p. 76–80, 2003.
[4]R.Salakhutdinov, A.Mnih, and G.Hinton, “Restricted Boltzmann Machines for Collaborative Filtering,” Proceedings of the 27th International Conference on Machine Learning, p. 791–798, 2007.
[5]S.Sedhain, A. K.Menon, S.Sanner, and L.Xie, “AutoRec : Autoencoders Meet Collaborative Filtering,” WWW 2015 Companion: Proceedings of the 24th International Conference on World Wide Web, p. 111–112, 2015.
[6]L.Zheng, V.Noroozi, and P. S.Yu, “Joint Deep Modeling of Users and Items Using Reviews for Recommendation,” Proceedings of the Tenth ACM International Conference on Web Search and Data Mining, p. 425–433, 2017.
[7]F.Yu, Q.Liu, S.Wu, L.Wang, and T.Tan, “A Dynamic Recurrent Model for Next Basket Recommendation,” Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval - SIGIR '16, p. 729–732, 2016.
[8]J.Li, P.Ren, Z.Chen, Z.Ren, and J.Ma, “Neural Attentive Session-based Recommendation,” Proceedings of the 2017 ACM on Conference on Information and Knowledge Management, 2017.
[9]Y.Song, A. M.Elkahky, and X.He, “Multi-Rate Deep Learning for Temporal Recommendation,” Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval - SIGIR '16, p. 909–912, 2016.
[10]P.-S.Huang, X.He, J.Gao, L.Deng, A.Acero, and L.Heck, “Learning deep structured semantic models for web search using clickthrough data,” Proceedings of the 22nd ACM international conference on Conference on information & knowledge management - CIKM '13, p. 2333–2338, 2013.
[11]L.Xiang et al., “Temporal recommendation on graphs via long- and short-term preference fusion,” Proceedings of the 16th ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '10, p. 723, 2010.
[12]L.Li, L.Zheng, F.Yang, and T.Li, “Modeling and broadening temporal user interest in personalized news recommendation,” Expert Systems with Applications, vol. 41, no. 7, p. 3168–3177, 2014.
[13]J.Chung, C.Gulcehre, K.Cho, and Y.Bengio, “Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling,” arXiv, p. 1–9, 2014.
[14]S.Hochreiter and J.Schmidhuber, “Long Short-Term Memory,” Neural Computation, vol. 9, no. 8, p.1735-1780, 1997.
[15]R.Collobert, J.Weston, L.Bottou, M.Karlen, K.Kavukcuoglu, and P.Kuksa, “Natural Language Processing (Almost) from Scratch,” Journal of Machine Learning Research, vol. 12, p. 2493–2537, 2011.
[16]V.Nair and G. E.Hinton, “Rectified Linear Units Improve Restricted Boltzmann Machines,” Proceedings of the 27th International Conference on Machine Learning, no. 3, p. 807–814, 2010.
[17]G.Hinton and T.Tieleman, “Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude,” COURSERA: Neural networks for machine learning, 2012.
[18]X.Geng, H.Zhang, J.Bian, and T. S.Chua, “Learning image and user features for recommendation in social networks,” Proceedings of the IEEE International Conference on Computer Vision, vol. 2015 Inter, p. 4274–4282, 2015.
[19]Y.Koren, R.Bell, and C.Volinsky, “Matrix Factorization Techniques for Recommender Systems,” Computer, vol. 42, no. 8, p. 30–37, 2009.
[20]S.Rendle, C.Freudenthaler, and L.Schmidt-Thieme, “Factorizing personalized Markov chains for next-basket recommendation,” Proceedings of the 19th international conference on World wide web - WWW '10, p. 811, 2010.
[21]D.Bahdanau, K.Cho, and Y.Bengio, “Neural Machine Translation by Jointly Learning to Align and Translate,” arXiv, p. 1–15, 2014.
論文全文使用權限
  • 同意授權校內瀏覽/列印電子全文服務,於2019-08-02起公開。


  • 如您有疑問,請聯絡圖書館
    聯絡電話:(06)2757575#65773
    聯絡E-mail:etds@email.ncku.edu.tw