透過超結構遷移的非線性跨領域協同過濾

Non-Linear Cross-Domain Collaborative Filtering via Hyper-Structure Transfer

劉彥甫(Liu, Yan Fu)

指導教授 ： 吳尚鴻

國立清華大學/電機資訊學院/資訊工程學系/碩士(2015年)

摘要

跨領域協同過濾利用來自多個領域的評分矩陣，以做出更好的推薦結果。現有的跨領域協同過濾方法採用子結構共享技術，只能在各領域之間傳送線性相關的資訊知識。在本文中，我們提出了超結構遷移的概念，即要求每個評分矩陣由一個所有領域共有的更複雜的結構的投影部分進行說明，稱為超結構，因此可以讓所有領域之間的非線性相關的知識得以鑑識和轉移。並以充實的實驗方式進行驗證，結果可以顯示我們的超領域遷移模型具有有效性。

關鍵字

協同過濾 ； 非線性

並列摘要

The Cross Domain Collaborative Filtering (CDCF) exploits the rating matrices from multiple domains to make better recommendations. Existing CDCF methods adopt the sub-structure sharing technique that can only transfer linearly correlated knowledge between domains. In this paper, we propose the notion of Hyper-Structure Transfer (HST) that requires the rating matrices to be explained by the projections of some more complex structure, called the hyper-structure, shared by all domains, and thus allows the non-linearly correlated knowledge between domains to be identified and transferred. Extensive experiments are conducted and the results demonstrate the effectiveness of our HST models empirically.

並列關鍵字

Collaborative Filtering ； Non-Linear

參考文獻

[1] Argyriou, Andreas, Andreas Maurer, and Massimiliano Pontil. An algorithm or transfer learning in a heterogeneous environment. In proc. of ECML PKDD, pages 71-85. ECML PKDD, 2008.

[4] Ben-David, S. and Schuller, R. Exploiting task relatedness for multiple task learning. In Proc. of COLT/Kernel, 2003.

[6] Chen, Wei, Hsu, Wynne, and Lee, Mong Li. Making recommendations from multiple domains. In Proc. of KDD, 2013.

[7] Ding, Chris, Li, Tao, Peng, Wei, and Park, Haesun. Orthogonal nonnegative matrix t-factorizations for clustering. In Proc. of KDD, 2006.

[11] Hofmann, Thomas. Latent semantic models for collaborative filtering. ACM Trans. on Information Systems, 22:89–115, 2004.