Factored Proximity Models for Top-N Recommendations

In this paper, we propose EIGENREC; a simple and versatile Latent Factor framework for Top-N Recommendations, which includes the well-known PureSVD algorithm as a special case. EIGENREC builds a low dimensional model of an inter-item proximity matrix that combines a traditional similarity component, with a scaling operator, designed to regulate the effects of the prior item popularity on the final recommendation list. A comprehensive set of experiments on the MovieLens and the Yahoo datasets, based on widely applied performance metrics suggest that EIGENREC outperforms several state of-the-art algorithms, in terms of Standard and Long-Tail recommendation accuracy, while exhibiting low susceptibility to the problems caused by Sparsity, even its most extreme manifestations – the Cold-start problems.

[1]  François Fouss,et al.  An experimental investigation of kernels on graphs for collaborative recommendation and semisupervised classification , 2012, Neural Networks.

[2]  Kenneth Y. Goldberg,et al.  Eigentaste: A Constant Time Collaborative Filtering Algorithm , 2001, Information Retrieval.

[3]  Yousef Saad,et al.  Lanczos Vectors versus Singular Vectors for Effective Dimension Reduction , 2009, IEEE Transactions on Knowledge and Data Engineering.

[4]  George Karypis,et al.  FISM: factored item similarity models for top-N recommender systems , 2013, KDD.

[5]  C. Lanczos.  An iteration method for the solution of the eigenvalue problem of linear differential and integral operators , 1950 .

[6]  R. Armstrong.  The Long Tail: Why the Future of Business Is Selling Less of More , 2008 .

[7]  Guy Shani,et al.  Evaluating Recommendation Systems , 2011, Recommender Systems Handbook.

[8]  Hsinchun Chen,et al.  Applying associative retrieval techniques to alleviate the sparsity problem in collaborative filtering , 2004, TOIS.

[9]  John D. Garofalakis,et al.  EigenRec: generalizing PureSVD for effective and efficient top-N recommendations , 2015, Knowledge and Information Systems.

[10]  John D. Garofalakis,et al.  NCDREC: A Decomposability Inspired Framework for Top-N Recommendation , 2014, 2014 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT).

[11]  Yousef Saad,et al.  Cucheb: A GPU implementation of the filtered Lanczos procedure , 2017, Comput. Phys. Commun..

[12]  George Karypis,et al.  A Comprehensive Survey of Neighborhood-based Recommendation Methods , 2011, Recommender Systems Handbook.

[13]  Domonkos Tikk,et al.  Scalable Collaborative Filtering Approaches for Large Recommender Systems , 2009, J. Mach. Learn. Res..

[14]  Roberto Turrin,et al.  Performance of recommender algorithms on top-n recommendation tasks , 2010, RecSys '10.

[15]  Yehuda Koren,et al.  Factorization meets the neighborhood: a multifaceted collaborative filtering model , 2008, KDD.

[16]  John D. Garofalakis,et al.  EIGENREC: An Efficient and Scalable Latent Factor Family for Top-N Recommendation , 2015, ArXiv.

[17]  John Riedl,et al.  Application of Dimensionality Reduction in Recommender System - A Case Study , 2000 .

[18]  Junjie Yao,et al.  Challenging the Long Tail Recommendation , 2012, Proc. VLDB Endow..

[19]  John D. Garofalakis,et al.  Hierarchical Itemspace Rank: Exploiting hierarchy to alleviate sparsity in ranking-based recommendation , 2015, Neurocomputing.

[20]  J. Bobadilla,et al.  Recommender systems survey , 2013, Knowl. Based Syst..

[21]  John D. Garofalakis,et al.  On the Use of Lanczos Vectors for Efficient Latent Factor-Based Top-N Recommendation , 2014, WIMS '14.

[22]  N NikolakopoulosAthanasios,et al.  Hierarchical Itemspace Rank , 2015 .