學術產出-Theses

Article View/Open

Publication Export

Google ScholarTM

政大圖書館

Citation Infomation

  • No doi shows Citation Infomation
題名 冷起始使用者的跨領域推薦機制
Cross-Domain Recommendation Mechanism for Cold-Start Users
作者 徐嘉泰
Hsu, Chia Tai
貢獻者 沈錳坤
Shan, Man Kwan
徐嘉泰
Hsu, Chia Tai
關鍵詞 推薦系統
跨領域推薦系統
冷起始使用者
偏最小平方法
偏最小平方迴歸
recommender system
cross-domain recommender system
cold-start user
partial least squares
partial least squares regression
日期 2016
上傳時間 2-Sep-2016 00:13:37 (UTC+8)
摘要 推薦系統在近代的網路購物中十分常見,在多數的電子商務網站皆可見其蹤跡。推薦系統可以幫助使用者在無垠的商品海中挑選適當的商品,節省使用者的時間與精力,同時也能夠提升商品的銷售量,增加商家的獲利。跨領域推薦系統(cross-domain recommender system)是推薦系統的其中一種,對於此方面的研究漸漸興起,跨領域推薦系統主要是希望利用使用者對不同領域物品的的喜好的特性進行推薦。
冷起始使用者問題(cold-start user problem)是推薦系統中十分常見的實務問題,在使用者缺乏資訊的情況下,推薦系統難以為這些使用者推薦物品、或是造成推薦的不準確性。目前有一些跨領域推薦系統的研究,但鮮少解決跨領域推薦中的冷起始使用者問題,特別是使用者在目標領域全無資訊的情況下的困境,因此本研究提出解決此問題的方法。
本研究旨在提出一個基於相關性分析的協同式過濾(collaborative filtering)跨領域推薦模型,該模型主要解決在多重領域的情況下使用者於目標領域的冷起始問題。本研究使用PLSR(Partial Least Squares Regression)作為整個模型的核心方法,利用PLSR迴歸預測的特性預測使用者跨領域的評分(rating)。此外,過程中結合其他的推薦技術以解決PLSR無法應用於有缺值(missing value)資料的問題。本研究最主要的貢獻,在於解決推薦對象在目標領域冷起始的情況低推薦準確率的問題,實驗也證明本研究所提出的推薦方法有不錯的Mean Absolute Error與Root Mean Squared Error評估數值,特別是在資料較少的情況下還能比其他推薦方法有更佳的推薦效果。
Recommender systems are common in the E-Commerce platforms in recent years. Recommender systems are able to help users find suitable items among large amount of products. It can save users’ time and increase profits of sellers. Cross-domain recommender systems, which related researches are blooming recently, is a sort of recommender systems that recommend items based on users’ different taste across domains.
Cold-start users problem is a common practical problem in recommender systems. In the case of scarce information of users, it is difficult for recommender systems to recommend items or even lead to inaccuracy of recommendation results. There are few studies on dealing with cold-start users problem in cross-domain recommender systems, especially in the case of shortage of information.
This thesis proposes a collaborative filtering cross-domain recommendation model based on correlation analysis, which mainly solve the cold-start problem that users have no preference information in the target domain in cross-domain situation. The proposed recommendation model is able to predict the rating among multiple domains for cold start users based on the Partial Least Squares Regression (PLSR). The proposed model incorporates other recommendation techniques for dealing with the problem that PLSR cannot use the data with missing values as input. The primary contribution of this research is that the proposed model is able to solve the problem of low recommendation accuracy for cold-start users in target domain. The experiment shows Mean Absolute Error and Root Mean Squared Error of the proposed model are better than the other approaches when data is sparse.
參考文獻 [1] G. Adomavicius and A. Tuzhilin, Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. IEEE Transactions on Knowledge and Data Engineering, 17(6), 734-749, 2005.
[2] J. Bobadilla, F. Ortega, A. Hernando, and A. Gutiérrez, Recommender systems survey. Knowledge-Based Systems, 46, 109-132, 2013.
[3] J. S. Breese, D. Heckerman, and C. Kadie, Empirical analysis of predictive algorithms for collaborative filtering. Proceedings of the 14th Conference on Uncertainty in Artificial Intelligence, 1998.
[4] I. Cantador, I. Fernández-Tobías, S. Berkovsky, and P. Cremonesi, Cross-domain recommender systems, Springer, 2015.
[5] P. Cremonesi, A. Tripodi, and R. Turrin, Cross-domain recommender systems. IEEE 11th International Conference on Data Mining Workshops, 2011.
[6] S. C. Deerwester, S. T. Dumais, T. K. Landauer, G. W. Furnas, and R. A. Harshman, Indexing by latent semantic analysis. Journal of the American Society For Information Science, 41(6), 391-407, 1990.
[7] I. Fernández-Tobías, I. Cantador, M. Kaminskas, and F. Ricci, Cross-domain recommender systems: A survey of the state of the art. Proceedings of the 2nd Spanish Conference on Information Retrieval, 2012.
[8] I. Jolliffe, Principal component analysis. Wiley Online Library, 2002.
[9] Y. Koren, R. Bell, and C. Volinsky, Matrix factorization techniques for recommender systems. Computer, 42(8), 30-37, 2009.
[10] D. D. Lee and H. S. Seung, Algorithms for non-negative matrix factorization. Advances in Neural Information Processing Systems, 2001.
[11] D. Li, N. Dimitrova, M. Li, and I. K. Sethi, Multimedia content processing through cross-modal association. Proceedings of the 11th ACM International Conference on Multimedia, 2003.
[12] B. Li, Q. Yang, and X. Xue, Can movies and books collaborate?: cross-domain collaborative filtering for sparsity reduction. Proceedings of the 21st International Joint Conference on Artifical Intelligence, 2009.
[13] R. Manne, Analysis of two partial-least-squares algorithms for multivariate calibration. Chemometrics and Intelligent Laboratory Systems, 2(1), 187-197, 1987.
[14] S. Meyffret, E. Guillot, L. Médini, and F. Laforest, RED: a Rich Epinions Dataset for Recommender Systems.2014.
[15] N. Mirbakhsh and C. X. Ling, Improving top-N recommendation for cold-start users via cross-domain information. ACM Transactions on Knowledge Discovery from Data, 9(4), 33, 2015.
[16] R. J. Mooney and L. Roy, Content-based book recommending using learning for text categorization. Proceedings of the 5th ACM conference on Digital Libraries, 2000.
[17] W. Pan, E. W. Xiang, N. N. Liu, and Q. Yang, Transfer learning in collaborative filtering for sparsity reduction. Proceedings of the 24th AAAI Conference on Artificial Intelligence, 2010.
[18] W. Pan, E. W. Xiang, and Q. Yang, Transfer learning in collaborative filtering with uncertain ratings. Proceedings of the 26th AAAI Conference on Artificial Intelligence, 2012.
[19] R. Rosipal and N. Krämer, Overview and recent advances in partial least squares. In Subspace, Latent Structure and Feature Selection, Springer, 34-51, 2006.
[20] B. Sarwar, G. Karypis, J. Konstan, and J. Riedl, Item-based collaborative filtering recommendation algorithms. Proceedings of the 10th International Conference on World Wide Web, 2001.
[21] B. Shapira, L. Rokach, and S. Freilikhman, Facebook single and cross domain data for recommendation systems. User Modeling and User-Adapted Interaction, 23(2-3), 211-247, 2013.
[22] Y. Shi, M. Larson, and A. Hanjalic, Tags as bridges between domains: improving recommendation with tag-induced cross-domain collaborative filtering. In User Modeling, Adaption and Personalization, Springer, 305-316, 2011.
[23] A. Tiroshi, S. Berkovsky, M. A. Kaafar, T. Chen, and T. Kuflik, Cross social networks interests predictions based ongraph features. Proceedings of the 7th ACM Conference on Recommender Systems, 2013.
[24] Y. Zhang, B. Cao, and D.-Y. Yeung, Multi-domain collaborative filtering. In Proceedings of the 26th Conference Annual Conference on Uncertainty in Artificial Intelligence, 2012
[25] F. Zhuang, P. Luo, H. Xiong, Y. Xiong, Q. He, and Z. Shi, Cross-domain learning from multiple sources: a consensus regularization perspective. IEEE Transactions on Knowledge and Data Engineering, 22(12), 1664-1678, 2010.
描述 碩士
國立政治大學
資訊科學學系
101753027
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0101753027
資料類型 thesis
dc.contributor.advisor 沈錳坤zh_TW
dc.contributor.advisor Shan, Man Kwanen_US
dc.contributor.author (Authors) 徐嘉泰zh_TW
dc.contributor.author (Authors) Hsu, Chia Taien_US
dc.creator (作者) 徐嘉泰zh_TW
dc.creator (作者) Hsu, Chia Taien_US
dc.date (日期) 2016en_US
dc.date.accessioned 2-Sep-2016 00:13:37 (UTC+8)-
dc.date.available 2-Sep-2016 00:13:37 (UTC+8)-
dc.date.issued (上傳時間) 2-Sep-2016 00:13:37 (UTC+8)-
dc.identifier (Other Identifiers) G0101753027en_US
dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/101129-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 資訊科學學系zh_TW
dc.description (描述) 101753027zh_TW
dc.description.abstract (摘要) 推薦系統在近代的網路購物中十分常見,在多數的電子商務網站皆可見其蹤跡。推薦系統可以幫助使用者在無垠的商品海中挑選適當的商品,節省使用者的時間與精力,同時也能夠提升商品的銷售量,增加商家的獲利。跨領域推薦系統(cross-domain recommender system)是推薦系統的其中一種,對於此方面的研究漸漸興起,跨領域推薦系統主要是希望利用使用者對不同領域物品的的喜好的特性進行推薦。
冷起始使用者問題(cold-start user problem)是推薦系統中十分常見的實務問題,在使用者缺乏資訊的情況下,推薦系統難以為這些使用者推薦物品、或是造成推薦的不準確性。目前有一些跨領域推薦系統的研究,但鮮少解決跨領域推薦中的冷起始使用者問題,特別是使用者在目標領域全無資訊的情況下的困境,因此本研究提出解決此問題的方法。
本研究旨在提出一個基於相關性分析的協同式過濾(collaborative filtering)跨領域推薦模型,該模型主要解決在多重領域的情況下使用者於目標領域的冷起始問題。本研究使用PLSR(Partial Least Squares Regression)作為整個模型的核心方法,利用PLSR迴歸預測的特性預測使用者跨領域的評分(rating)。此外,過程中結合其他的推薦技術以解決PLSR無法應用於有缺值(missing value)資料的問題。本研究最主要的貢獻,在於解決推薦對象在目標領域冷起始的情況低推薦準確率的問題,實驗也證明本研究所提出的推薦方法有不錯的Mean Absolute Error與Root Mean Squared Error評估數值,特別是在資料較少的情況下還能比其他推薦方法有更佳的推薦效果。
zh_TW
dc.description.abstract (摘要) Recommender systems are common in the E-Commerce platforms in recent years. Recommender systems are able to help users find suitable items among large amount of products. It can save users’ time and increase profits of sellers. Cross-domain recommender systems, which related researches are blooming recently, is a sort of recommender systems that recommend items based on users’ different taste across domains.
Cold-start users problem is a common practical problem in recommender systems. In the case of scarce information of users, it is difficult for recommender systems to recommend items or even lead to inaccuracy of recommendation results. There are few studies on dealing with cold-start users problem in cross-domain recommender systems, especially in the case of shortage of information.
This thesis proposes a collaborative filtering cross-domain recommendation model based on correlation analysis, which mainly solve the cold-start problem that users have no preference information in the target domain in cross-domain situation. The proposed recommendation model is able to predict the rating among multiple domains for cold start users based on the Partial Least Squares Regression (PLSR). The proposed model incorporates other recommendation techniques for dealing with the problem that PLSR cannot use the data with missing values as input. The primary contribution of this research is that the proposed model is able to solve the problem of low recommendation accuracy for cold-start users in target domain. The experiment shows Mean Absolute Error and Root Mean Squared Error of the proposed model are better than the other approaches when data is sparse.
en_US
dc.description.tableofcontents 第一章 緒論 1
第二章 相關研究 4
第三章 研究方法 10
3.1 推薦目標與推薦情境 10
3.2 系統架構:PLSR_CrossRec 11
3.3 矩陣分解填值 12
3.4 Partial Least Squares Regression 14
3.5 整體推薦架構:PLSR_CrossRec 21
3.6 更有效率的方法:PLSR_Latent 22
第四章 效能評估 24
4.1 資料集 24
4.2 資料切割方法 24
4.3 評估方法 25
4.4 比較推薦方法 26
4.5 實驗設計與結果 29
4.5.1 同質性的來源領域與目標領域 30
4.5.2 異質性的來源領域與目標領域 31
4.5.3 複合性質的來源領域 33
4.5.4 CrossCBMF之 變化對於mae與rmse之影響 34
4.5.5 矩陣密度對於推薦效果的影響 36
4.5.6 PLSR不同k值運算時間 38
第五章 結論與未來研究 40
參考文獻 41
zh_TW
dc.format.extent 1116518 bytes-
dc.format.mimetype application/pdf-
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0101753027en_US
dc.subject (關鍵詞) 推薦系統zh_TW
dc.subject (關鍵詞) 跨領域推薦系統zh_TW
dc.subject (關鍵詞) 冷起始使用者zh_TW
dc.subject (關鍵詞) 偏最小平方法zh_TW
dc.subject (關鍵詞) 偏最小平方迴歸zh_TW
dc.subject (關鍵詞) recommender systemen_US
dc.subject (關鍵詞) cross-domain recommender systemen_US
dc.subject (關鍵詞) cold-start useren_US
dc.subject (關鍵詞) partial least squaresen_US
dc.subject (關鍵詞) partial least squares regressionen_US
dc.title (題名) 冷起始使用者的跨領域推薦機制zh_TW
dc.title (題名) Cross-Domain Recommendation Mechanism for Cold-Start Usersen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) [1] G. Adomavicius and A. Tuzhilin, Toward the next generation of recommender systems: A survey of the state-of-the-art and possible extensions. IEEE Transactions on Knowledge and Data Engineering, 17(6), 734-749, 2005.
[2] J. Bobadilla, F. Ortega, A. Hernando, and A. Gutiérrez, Recommender systems survey. Knowledge-Based Systems, 46, 109-132, 2013.
[3] J. S. Breese, D. Heckerman, and C. Kadie, Empirical analysis of predictive algorithms for collaborative filtering. Proceedings of the 14th Conference on Uncertainty in Artificial Intelligence, 1998.
[4] I. Cantador, I. Fernández-Tobías, S. Berkovsky, and P. Cremonesi, Cross-domain recommender systems, Springer, 2015.
[5] P. Cremonesi, A. Tripodi, and R. Turrin, Cross-domain recommender systems. IEEE 11th International Conference on Data Mining Workshops, 2011.
[6] S. C. Deerwester, S. T. Dumais, T. K. Landauer, G. W. Furnas, and R. A. Harshman, Indexing by latent semantic analysis. Journal of the American Society For Information Science, 41(6), 391-407, 1990.
[7] I. Fernández-Tobías, I. Cantador, M. Kaminskas, and F. Ricci, Cross-domain recommender systems: A survey of the state of the art. Proceedings of the 2nd Spanish Conference on Information Retrieval, 2012.
[8] I. Jolliffe, Principal component analysis. Wiley Online Library, 2002.
[9] Y. Koren, R. Bell, and C. Volinsky, Matrix factorization techniques for recommender systems. Computer, 42(8), 30-37, 2009.
[10] D. D. Lee and H. S. Seung, Algorithms for non-negative matrix factorization. Advances in Neural Information Processing Systems, 2001.
[11] D. Li, N. Dimitrova, M. Li, and I. K. Sethi, Multimedia content processing through cross-modal association. Proceedings of the 11th ACM International Conference on Multimedia, 2003.
[12] B. Li, Q. Yang, and X. Xue, Can movies and books collaborate?: cross-domain collaborative filtering for sparsity reduction. Proceedings of the 21st International Joint Conference on Artifical Intelligence, 2009.
[13] R. Manne, Analysis of two partial-least-squares algorithms for multivariate calibration. Chemometrics and Intelligent Laboratory Systems, 2(1), 187-197, 1987.
[14] S. Meyffret, E. Guillot, L. Médini, and F. Laforest, RED: a Rich Epinions Dataset for Recommender Systems.2014.
[15] N. Mirbakhsh and C. X. Ling, Improving top-N recommendation for cold-start users via cross-domain information. ACM Transactions on Knowledge Discovery from Data, 9(4), 33, 2015.
[16] R. J. Mooney and L. Roy, Content-based book recommending using learning for text categorization. Proceedings of the 5th ACM conference on Digital Libraries, 2000.
[17] W. Pan, E. W. Xiang, N. N. Liu, and Q. Yang, Transfer learning in collaborative filtering for sparsity reduction. Proceedings of the 24th AAAI Conference on Artificial Intelligence, 2010.
[18] W. Pan, E. W. Xiang, and Q. Yang, Transfer learning in collaborative filtering with uncertain ratings. Proceedings of the 26th AAAI Conference on Artificial Intelligence, 2012.
[19] R. Rosipal and N. Krämer, Overview and recent advances in partial least squares. In Subspace, Latent Structure and Feature Selection, Springer, 34-51, 2006.
[20] B. Sarwar, G. Karypis, J. Konstan, and J. Riedl, Item-based collaborative filtering recommendation algorithms. Proceedings of the 10th International Conference on World Wide Web, 2001.
[21] B. Shapira, L. Rokach, and S. Freilikhman, Facebook single and cross domain data for recommendation systems. User Modeling and User-Adapted Interaction, 23(2-3), 211-247, 2013.
[22] Y. Shi, M. Larson, and A. Hanjalic, Tags as bridges between domains: improving recommendation with tag-induced cross-domain collaborative filtering. In User Modeling, Adaption and Personalization, Springer, 305-316, 2011.
[23] A. Tiroshi, S. Berkovsky, M. A. Kaafar, T. Chen, and T. Kuflik, Cross social networks interests predictions based ongraph features. Proceedings of the 7th ACM Conference on Recommender Systems, 2013.
[24] Y. Zhang, B. Cao, and D.-Y. Yeung, Multi-domain collaborative filtering. In Proceedings of the 26th Conference Annual Conference on Uncertainty in Artificial Intelligence, 2012
[25] F. Zhuang, P. Luo, H. Xiong, Y. Xiong, Q. He, and Z. Shi, Cross-domain learning from multiple sources: a consensus regularization perspective. IEEE Transactions on Knowledge and Data Engineering, 22(12), 1664-1678, 2010.
zh_TW