Manolis G. Vozalis

Using SVD and demographic data for the enhancement of generalized Collaborative Filtering

In this paper we examine how Singular Value Decomposition (SVD) along with demographic information can enhance plain Collaborative Filtering (CF) algorithms. After a brief introduction to SVD, where some of its previous applications in Recommender Systems are revisited, we proceed with a full description of our proposed method which utilizes SVD and demographic data at various points of the filtering procedure in order to improve the quality of the generated predictions. We test the efficiency of the resulting approach on two commonly used CF approaches (User-based and Item-based CF). The experimental part of this work involves a number of variations of the proposed approach. The results show that the combined utilization of SVD with demographic data is promising, since it does not only tackle some of the recorded problems of Recommender Systems, but also assists in increasing the accuracy of systems employing it.

Applying SVD on item-based filtering

In this paper we examine the use of a matrix factorization technique called singular value decomposition (SVD) in item-based collaborative filtering. After a brief introduction to SVD and some of its previous applications in recommender systems, we proceed with a full description of our algorithm, which uses SVD in order to reduce the dimension of the active items neighborhood. The experimental part of this work first locates the ideal parameter settings for the algorithm, and concludes by contrasting it with plain item-based filtering which utilizes the original, high dimensional neighborhood. The results show that a reduction in the dimension of the item neighborhood is promising, since it does not only tackle some of the recorded problems of recommender systems, but also assists in increasing the accuracy of systems employing it.

Collaborative filtering through SVD-based and hierarchical nonlinear PCA

In this paper, we describe and compare two distinct algorithms aiming at the low-rank approximation of a user-item ratings matrix in the context of Collaborative Filtering (CF). The first one implements standard Principal Component Analysis (PCA) of an association matrix formed from the original data. The second algorithm is based on h-NLPCA, a nonlinear generalization of standard PCA, which utilizes an autoassociative network, and constrains the nonlinear components to have the same hierarchical order as the linear components in standard PCA. We examine the impact of the aforementioned approaches on the quality of the generated predictions through a series of experiments. Experimental results show that the latter approach outperforms the standard PCA approach for most values of the retained dimensions.

Unison-CF: A multiple-component, adaptive collaborative filtering system

In this paper we present the Unison-CF algorithm, which provides an efficient way to combine multiple collaborative filtering approaches, drawing advantages from each one of them. Each collaborative filtering approach is treated as a separate component, allowing the Unison-CF algorithm to be easily extended. We evaluate the Unison-CF algorithm by applying it on three existing filtering approaches: User-based Filtering, Item-based Filtering and Hybrid-CF. Adaptation is utilized and evaluated as part of the filtering approaches combination. Our experiments show that the Unison-CF algorithm generates promising results in improving the accuracy and coverage of the existing filtering algorithms.

On the combination of user-based and item-based collaborative filtering

In this paper, we propose two new filtering algorithms which are a combination of user-based and item-based collaborative filtering schemes. The first one, Hybrid-Ib, identifies a reasonably large neighbourhood of similar users and then uses this subset to derive the item-based recommendation model. The second algorithm, Hybrid-CF, starts by locating items similar to the one for which we want a prediction, and then, based on that neighbourhood, it generates its user-based predictions. We start by describing the execution steps of the algorithms and proceed with extended experiments. We conclude that our algorithms are directly comparable to existing filtering approaches, with Hybrid-CF producing favorable or, in the worst case, similar results in all selected evaluation metrics. E-mail: kmarg@uom.gr

An Optimal Scaling Approach to Collaborative Filtering Using Categorical Principal Component Analysis and Neighborhood Formation

Collaborative Filtering (CF) is a popular technique employed by Recommender Systems, a term used to describe intelligent methods that generate personalized recommendations. The most common and accurate approaches to CF are based on latent factor models. Latent factor models can tackle two fundamental problems of CF, data sparsity and scalability and have received considerable attention in recent literature. In this work, we present an optimal scaling approach to address both of these problems using Categorical Principal Component Analysis for the low-rank approximation of the user-item ratings matrix, followed by a neighborhood formation step. The optimal scaling approach has the advantage that it can be easily extended to the case when there are missing data and restrictions for ordinal and numerical variables can be easily imposed. We considered different measurement levels for the user ratings on items, starting with a multiple nominal and consecutively applying nominal, ordinal and numeric levels. Experiments were executed on the MovieLens dataset, aiming to evaluate the aforementioned options in terms of accuracy. Results indicated that a combined approach (multiple nominal measurement level, ‘‘passive’’ missing data strategy) clearly outperformed the other tested options.

Identifying the effects of SVD and demographic data use on generalized collaborative filtering

The purpose of this paper is to examine how singular value decomposition (SVD) and demographic information can improve the performance of plain collaborative filtering (CF) algorithms. After a brief introduction to SVD, where the method is explained and some of its applications in recommender systems are detailed, we focus on the proposed technique. Our approach applies SVD in different stages of an algorithm, which can be described as CF enhanced by demographic data. The results of a rather long series of experiments, where the proposed algorithm is successfully blended with user- and item-based CF, show that the combined utilization of SVD with demographic data is promising, since it does not only tackle some of the recorded problems of recommender systems, but also assists in increasing the accuracy of systems employing it.

On the Performance of SVD-Based Algorithms for Collaborative Filtering

In this paper, we describe and compare threeCollaborative Filtering (CF) algorithms aiming at the low-rank approximation of the user-item ratings matrix. The algorithm implementations are based on three standard techniques for fitting a factor model to the data: Standard Singular Value Decomposition (sSVD), Principal Component Analysis (PCA) and Correspondence Analysis (CA). CA and PCA can be described as SVDs of appropriately transformed matrices,which is a key concept in this study. For each algorithm we implement two similar CF versions. The first one involves a direct rating prediction scheme based on the reduced user-item ratings matrix, while the second incorporates an additional neighborhood formation step. Next, we examine the impact of the aforementioned approaches on the quality of the generated predictions through a series of experiments. The experimental results showed that the approaches including the neighborhood formation step in most cases appear to be less accurate thanthe direct ones. Finally, CA-CF outperformed the SVD-CFand PCA-CF in terms of accuracy for small numbers ofretained dimensions, but SVD-CF displayed the overall highest accuracy.

AN OPTIMAL SCALING FRAMEWORK FOR COLLABORATIVE FILTERING RECOMMENDATION SYSTEMS

Collaborative Filtering (CF) is a popular technique employed by Recommender Systems, a term used to describe intelligent methods that generate personalized recommendations. Some of the most efficie...

A Hybrid Approach for Improving Prediction Coverage of Collaborative Filtering

In this paper we present a hybrid filtering algorithm that attempts to deal with low prediction Coverage, a problem especially present in sparse datasets. We focus on Item HyCoV, an implementation of the proposed approach that incorporates an additional User-based step to the base Item-based algorithm, in order to take into account the possible contribution of users similar to the active user. A series of experiments were executed, aiming to evaluate the proposed approach in terms of Coverage and Accuracy. The results show that Item HyCov significantly improves both performance measures, requiring no additional data and minimal modification of existing filtering systems.