EIGENREC: An Efficient and Scalable Latent Factor Family for Top-N Recommendation

  title={EIGENREC: An Efficient and Scalable Latent Factor Family for Top-N Recommendation},
  author={Athanasios N. Nikolakopoulos and Vassilis Kalantzis and John D. Garofalakis},
Sparsity presents one of the major challenges of Collaborative Filtering. Graph-based methods are known to alleviate its effects, however their use is often computationally prohibitive; Latent-Factor methods, on the other hand, present a reasonable and viable alternative. In this paper, we introduce EigenRec; a versatile and efficient Latent-Factor framework for Top-N Recommendations, that generalizes the well-known PureSVD algorithm (a) providing intuition about its inner structure, (b) paving… Expand
LLFR: A Lanczos-Based Latent Factor Recommender for Big Data Scenarios
Lanczos Latent Factor Recommender (LLFR) is proposed; a novel "big data friendly" collaborative filtering algorithm for top-N recommendation that builds a low dimensional item similarity model, that can be readily exploited to produce personalized ranking vectors over the item space. Expand
Factored Proximity Models for Top-N Recommendations
It is suggested 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. Expand
HybridSVD: when collaborative information is not enough
A new hybrid algorithm that allows incorporating both user and item side information within the standard collaborative filtering technique and shows its superiority over similar classes of hybrid models. Expand
Ranking under near decomposability
In this dissertation, we study the problem of Ranking in the presence of Sparsity focusing on two of the most important and generic ranking settings; namely Link Analysis and Top-N Recommendation.Expand
Improving performances of Top-N recommendations with co-clustering method
A new recommendation method based on collaborative filtering called User-Item Community Detection based Recommendation (UICDR) method, modified from the previous work, that significantly improves the performances of Top-N recommendations of several traditional collaborative filtering methods. Expand
Bayesian pairwise learning to rank via one-class collaborative filtering
A novel collaborative pairwise learning to rank method referred to as BPLR is proposed, which aims to improve the performance of personalized ranking from implicit feedback, and takes account of the neighborhood relationship between users as well as the item similarity while deriving the potential candidates. Expand
A multistep priority-based ranking for top-N recommendation using social and tag information
The results are compared with that of the obtained from some state-of-the-art ranking methods and observe that recommendation accuracy is improved in the case of the proposed algorithm for both all users and cold-start users scenarios. Expand
Cucheb: A GPU implementation of the filtered Lanczos procedure
The software package Cucheb is described, a GPU implementation of the filtered Lanczos procedure for the solution of large sparse symmetric eigenvalue problems, and it is shown that using the GPU can reduce the computation time by more than a factor of 10. Expand
Boosting Item-based Collaborative Filtering via Nearly Uncoupled Random Walks
This work has shown that random walks can provide a powerful tool for harvesting the rich network of interac... Expand


On the Use of Lanczos Vectors for Efficient Latent Factor-Based Top-N Recommendation
A number of experiments indicate that LLFR outperforms other state-of-the-art top-N recommendation methods from a computational as well as a qualitative perspective, and its relative performance gains increase as the data get sparser. Expand
Top-N recommendations in the presence of sparsity: An NCD-based approach
This work exploits the innately hierarchical structure of the item space following an approach inspired by the theory of Decomposability to derive sufficient conditions that guarantee full item space coverage even in cold-start recommendation scenarios. Expand
NCDREC: A Decomposability Inspired Framework for Top-N Recommendation
A comprehensive set of experiments support the model's theoretically predicted properties and verify that NCDREC outperforms several state-of-the-art algorithms, in terms of recommendation accuracy, diversity and sparseness insensitivity. Expand
Hierarchical Itemspace Rank: Exploiting hierarchy to alleviate sparsity in ranking-based recommendation
HIR is a novel recommendation algorithm that exploits the intrinsic hierarchical structure of the itemspace to tackle sparsity, and to alleviate the related limitations it imposes to the quality of recommendation. Expand
Scalable Collaborative Filtering Approaches for Large Recommender Systems
This work proposes various scalable solutions that are validated against the Netflix Prize data set, currently the largest publicly available collection of CF techniques, and proposes various matrix factorization (MF) based techniques. Expand
FISM: factored item similarity models for top-N recommender systems
An item-based method for generating top-N recommendations that learns the item-item similarity matrix as the product of two low dimensional latent factor matrices using a structural equation modeling approach. Expand
Factorization meets the neighborhood: a multifaceted collaborative filtering model
The factor and neighborhood models can now be smoothly merged, thereby building a more accurate combined model and a new evaluation metric is suggested, which highlights the differences among methods, based on their performance at a top-K recommendation task. Expand
OrdRec: an ordinal model for predicting personalized item rating distributions
A collaborative filtering recommendation framework, which is based on viewing user feedback on products as ordinal, rather than the more common numerical view, that is more principled and empirically superior in its accuracy. Expand
Learning Attribute-to-Feature Mappings for Cold-Start Recommendations
This work uses the mapping concept to construct an attribute-aware matrix factorization model for item recommendation from implicit, positive-only feedback, and shows that this approach provides good predictive accuracy, while the prediction time only grows by a constant factor. Expand
Collaborative ranking
This work proposes novel models which approximately optimize NDCG for the recommendation task, essentially variations on matrix factorization models where the features associated with the users and the items for the ranking task are learned. Expand