本文中，我們提出了一個基於項目屬性的協同過濾推薦系統，它結合了項目屬性、調合平均權重、分群來推薦項目給用戶。我們的系統使用項目屬性來描述用戶的偏好，以便我們緩解資料稀疏與冷啟動的問題，並做出更廣泛的預測。我們使用調合平均權和分群來調整傳統的預測函數，因為我們認為用戶是否在同一群中，以及用戶的評分次數、共同評分次數，都會與預測結果有關，調合平均權重由兩個用戶間的評分次數、共同評分次數來計算。我們利用CIM當中的分群法對用戶進行分群，分群的結果與調合平均權重分別是為一個權重來調整預測函數。我們的方法可以在用戶間沒有共同評分的項目時也可以計算相似度，這是傳統協同過濾系統所做不到的。

In this article, we propose a tag-based collaborative filtering recommendation system which combines items’ tags, harmonic mean weight and cluster to recommend items to user. Our system uses items’ tags to describe user’s preference, so that we ease up data sparsity problem and cold start problem and make more widely prediction. We use harmonic mean weight and clustering to improve the former prediction function because we think that whether user are in the same cluster or not, and rating, co-rating times are related to predicted results. Harmonic mean weight[11] is calculated by rating times of two user. We utilize clustering method in CIM[12] to group user and see cluster result as a weight in prediction function. Our method can make prediction in some situation that other method cannot predict.

Table of Contents

Chinese Abstract……………………………………………………….. I
Abstract……………………………………………………………… II
Table of Contents…………………………………………………….. III
List of Figures……………………………………………………….. V
List of Tables……………………………………………………….. VI
Chapter 1 1
Introduction 1
Chapter 2 5
Related Work 5
2.1 User-Based Collaborative Filtering 5
2.2 Item-Based Collaborative Filtering 5
2.3 Model-Based Collaborative Filtering 6
2.4 Hybrid Method 7
Chapter 3 9
Tag-based Collaborative Filtering 9
3.1 User Preference and Normalize 10
3.2 User Similarity 11
3.3 H-Clustering in CIM 12
3.4 Prediction Function 16
Chapter 4 17
Experiments 17
4.1 Data Set 17
4.2 Evaluation Metric 17
4.3 Experiment Results 18
Chapter 5 21
Conclusion 21
Reference 22

























List of Figures

Figure. 1: System Architecture 9
Figure. 2: An example to show community discovery by H Clustering.. 14
Figure. 3: Difference of MAE in 100K rating data. 19
Figure. 4: Difference of RMSE in 100K rating data 19
Figure. 5: Difference of MAE in 1M rating data 19
Figure. 6: Difference of RMSE in 1M rating data. 20


List of Table

Table. 1: Example user-item matrix 10
Table. 2: Example user-tag matrix 11
Table. 3: MovieLens data in 100K and 1M 17
Table. 4: MAE of tag-based in different similarity method 18
Table. 5: RMSE of tag-based in different similarity method 18

[1]Kabbur, Santosh, and George Karypis. "Nlmf: Nonlinear matrix factorization methods for top-n recommender systems." 2014 IEEE International Conference on Data Mining Workshop. IEEE, 2014. (TOP-N)
[2]Gupta, Jyoti, and Jayant Gadge. "Performance analysis of recommendation system based on collaborative filtering and demographics." Communication, Information & Computing Technology (ICCICT), 2015 International Conference on. IEEE, 2015. (UB+IB)
[3]Kumar, Anuranjan, et al. "Comparison of various metrics used in collaborative filtering for recommendation system." Contemporary Computing (IC3), 2015 Eighth International Conference on. IEEE, 2015.(測量指標)
[4]Ma, Zhaocai, et al. "The SOM Based Improved K-Means Clustering Collaborative Filtering Algorithm in TV Recommendation System." Advanced Cloud and Big Data (CBD), 2014 Second International Conference on. IEEE, 2014. (UBCF)
[5]Jiang, Shuhui, et al. "Author topic model-based collaborative filtering for personalized POI recommendations." IEEE Transactions on Multimedia 17.6 (2015): 907-918. (modele-based)
[6]Ba, Qilong, Xiaoyong Li, and Zhongying Bai. "Clustering collaborative filtering recommendation system based on SVD algorithm." Software Engineering and Service Science (ICSESS), 2013 4th IEEE International Conference on. IEEE, 2013. (MODEL-based)
[7]Renaud-Deputter, Simon, Tengke Xiong, and Shengrui Wang. "Combining collaborative filtering and clustering for implicit recommender system." Advanced Information Networking and Applications (AINA), 2013 IEEE 27th International Conference on. IEEE, 2013. (KNN clustering in CF)
[8]Hofmann, Thomas. "Collaborative filtering via gaussian probabilistic latent semantic analysis." Proceedings of the 26th annual international ACM SIGIR conference on Research and development in information retrieval. ACM, 2003. (model-based )
[9]Pirasteh, Parivash, Dosam Hwang, and Jason J. Jung. "Exploiting matrix factorization to asymmetric user similarities in recommendation systems." Knowledge-Based Systems 83 (2015): 51-57.(matrix factorization)
[10]Niemann, Katja, and Martin Wolpers. "A new collaborative filtering approach for increasing the aggregate diversity of recommender systems." Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2013. (new CF) -> item-based
[11]Melville, Prem, Raymond J. Mooney, and Ramadass Nagarajan. "Content-boosted collaborative filtering for improved recommendations." Aaai/iaai. 2002. (content-based + UBCF)
[12]Chen, Yi-Cheng, et al. "CIM: community-based influence maximization in social networks." ACM Transactions on Intelligent Systems and Technology (TIST) 5.2 (2014): 25.
[13]Zhang, Ziyang, et al. "Selecting influential and trustworthy neighbors for collaborative filtering recommender systems." Computing and Communication Workshop and Conference (CCWC), 2017 IEEE 7th Annual. IEEE, 2017. (UBCF 改良)
[14]Wang, Jing, and Jian Yin. "Combining user-based and item-based collaborative filtering techniques to improve recommendation diversity." 2013 6th International Conference on Biomedical Engineering and Informatics. IEEE, 2013.
[15]Zhou, Weixevg, et al. "A collaborative filtering algorithm based on biclustering." Machine Learning and Cybernetics (ICMLC), 2015 International Conference on. Vol. 2. IEEE, 2015.
[16]Cai, Yi, et al. "Typicality-based collaborative filtering recommendation." IEEE Transactions on Knowledge and Data Engineering 26.3 (2014): 766-779.
[17]Nie, YanPing, Yang Liu, and Xiaohui Yu. "Weighted aspect-based collaborative filtering." Proceedings of the 37th international ACM SIGIR conference on Research & development in information retrieval. ACM, 2014.
[18]Chen, Gang, Fei Wang, and Changshui Zhang. "Collaborative filtering using orthogonal nonnegative matrix tri-factorization." Information Processing & Management 45.3 (2009): 368-379.
[19]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.
[20]Koren, Yehuda, Robert Bell, and Chris Volinsky. "Matrix factorization techniques for recommender systems." Computer 42.8 (2009): 30-37