Comparing Preference Models in Recommender Systems

How to represent the users' preference is one of the principle problems in widely used recommender systems. To address this problem, in this paper, the expressibility, space complexity and learning complexity of different kinds of preference models are investigated. The recommendation performances of these models are also compared on a real-life dataset. Considering both the expressibility and computational complexity, the quadric model is the most suited to recommender systems.

[1]  Yi Xiong,et al.  Learning Conditional Preference Networks from Inconsistent Examples , 2014, IEEE Transactions on Knowledge and Data Engineering.

[2]  Craig Boutilier,et al.  CP-nets: a tool for represent-ing and reasoning with conditional ceteris paribus state-ments , 2004 .

[3]  Ronen I. Brafman,et al.  UCP-Networks: A Directed Graphical Representation of Conditional Utilities , 2001, UAI.

[4]  Juntao Liu,et al.  Bayesian Probabilistic Matrix Factorization with Social Relations and Item Contents for recommendation , 2013, Decis. Support Syst..

[5]  Arkadiusz Paterek,et al.  Improving regularized singular value decomposition for collaborative filtering , 2007 .

[6]  Juntao Liu,et al.  Conditional preference in recommender systems , 2015, Expert Syst. Appl..

[7]  Loizos Michael,et al.  Ceteris Paribus Preference Elicitation with Predictive Guarantees , 2009, IJCAI.

[8]  Yi Xiong,et al.  Learning conditional preference network from noisy samples using hypothesis testing , 2013, Knowl. Based Syst..

[9]  Jaana Kekäläinen,et al.  IR evaluation methods for retrieving highly relevant documents , 2000, SIGIR '00.

[10]  Jérôme Lang,et al.  The Complexity of Learning Separable ceteris paribus Preferences , 2009, IJCAI.

[11]  Yi Xiong,et al.  List-wise probabilistic matrix factorization for recommendation , 2014, Inf. Sci..

[12]  Jaana Kekäläinen,et al.  IR evaluation methods for retrieving highly relevant documents , 2000, SIGIR Forum.

[13]  Ruslan Salakhutdinov,et al.  Probabilistic Matrix Factorization , 2007, NIPS.

[14]  Ronen I. Brafman,et al.  Reasoning With Conditional Ceteris Paribus Preference Statements , 1999, UAI.

[15]  Sean M. McNee,et al.  Improving recommendation lists through topic diversification , 2005, WWW '05.

[16]  Olivier Chapelle,et al.  Expected reciprocal rank for graded relevance , 2009, CIKM.

[17]  Junwei Wang,et al.  Representing conditional preference by boosted regression trees for recommendation , 2016, Inf. Sci..