Learning Context-Aware Ranking

In this chapter, we propose a learning method for the problem setting of contextaware ranking. This problem setting has been investigated in detail in the last chapter. We have seen, that a context-aware ranking Open image in new window can be modelled by a real-valued function Open image in new window . Now, we will show how this function can be optimized. The optimization will be done with respect to the pairwise training data d s , that is inferred from the sparse and incomplete observations s. The whole chapter assumes, that y can be expressed as a differentiable, non-recursive function with a finite set of parameters Θ. This assumption holds for many models, including the factorization models that we will introduce in the next chapter.

[1]  Joos Vandewalle,et al.  A Multilinear Singular Value Decomposition , 2000, SIAM J. Matrix Anal. Appl..

[2]  Yifan Hu,et al.  Collaborative Filtering for Implicit Feedback Datasets , 2008, 2008 Eighth IEEE International Conference on Data Mining.

[3]  Rong Pan,et al.  Mind the gaps: weighting the unknown in large-scale one-class collaborative filtering , 2009, KDD.

[4]  Tommi S. Jaakkola,et al.  Maximum-Margin Matrix Factorization , 2004, NIPS.

[5]  Qiang Yang,et al.  One-Class Collaborative Filtering , 2008, 2008 Eighth IEEE International Conference on Data Mining.

[6]  Panagiotis Symeonidis,et al.  Tag recommendations based on tensor dimensionality reduction , 2008, RecSys '08.

[7]  Bhavani Raskutti,et al.  Optimising area under the ROC curve using gradient descent , 2004, ICML.