Privacy-preserving Collaborative Filtering Research Articles

A novel target item-based similarity function in privacy-preserving collaborative filtering

Memory-based collaborative filtering schemes are among the most effective recommendation technologies in terms of prediction quality, despite commonly facing issues related to accuracy, scalability, and privacy. A prominent approach suggests an intuitively reasonable modification to the similarity function, which has been proven to provide more accurate recommendations than those generated by state-of-the-art memory-based collaborative filtering methods. However, this scheme exacerbates the scalability problem due to additional computational costs and fails to protect individual privacy. In this study, we recommend using a preprocessing method to eliminate relatively dissimilar items from the prediction estimation process, thereby enhancing the scalability of the proposed approach. We explore how to provide recommendations based on the previously proposed similarity function while preserving privacy and propose privacy-preserving schemes to accomplish this task. Additionally, we apply our preprocessing approach to our proposed privacy-preserving schemes to improve both scalability and accuracy. After analyzing our schemes with respect to privacy and additional costs, we conduct experiments with real data to examine the impact of our schemes on scalability and accuracy. The empirical outcomes indicate that our preprocessing scheme significantly alleviates scalability issues in both conventional and privacy-preserving environments and enhances accuracy within privacy-preserving frameworks.

Privacy-preserving Collaborative Filtering by Distributed Mediation

Recommender systems have become very influential in our everyday decision making, e.g., helping us choose a movie from a content platform, or offering us suitable products on e-commerce websites. While most vendors who utilize recommender systems rely exclusively on training data consisting of past transactions that took place through them, it would be beneficial to base recommendations on the rating data of more than one vendor. However, enlarging the training data by means of sharing information between different vendors may jeopardize the privacy of users. We devise here secure multi-party protocols that enable the practice of Collaborative Filtering (CF) in a manner that preserves the privacy of the vendors and users. Shmueli and Tassa [ 38 ] introduced privacy-preserving protocols of CF that involved a mediator; namely, an external entity that assists in performing the computations. They demonstrated the significant advantages of mediation in that context. We take here the mediation approach into the next level by using several independent mediators. Such distributed mediation maintains all of the advantages that were identified by Shmueli and Tassa, and offers additional ones, in comparison with the single-mediator protocols: stronger security and dramatically shorter runtimes. In addition, while all prior art assumed limited and unrealistic settings, in which each user can purchase any given item through only one vendor, we consider here a general and more realistic setting, which encompasses all previously considered settings, where users can choose between different competing vendors. We demonstrate the appealing performance of our protocols through extensive experimentation.

Privacy-preserving collaborative filtering: A new approach based on variable-group-size microaggregation

Abstract The amount of available information is growing steadily and, as a result, Internet users benefit from recommender systems, which help them find information, services and products that best fit their needs. A common technique used in recommender systems is Collaborative Filtering, which is based on users’ collaboration to make recommendations. However, users are getting more concerned about their privacy and can be reluctant to disclose their interests and other personal information. Hence, with the aim to foster users collaboration, the development of privacy-aware collaborative filtering methods has become a hot topic in the field. In this article we recall the concept of Privacy-Preserving Collaborative Filtering (PPCF) and introduce a novel approach based on variable-group-size microaggregation, which provides k-anonymity to the users. Also, we introduce several new metrics based on users’ behaviour that overcome the drawbacks of traditional metrics. Extensive experiments show that our approach can provide more accurate recommendations than well-known methods while, at the same time, preserving users’ privacy.

Improving k-anonymity based privacy preservation for collaborative filtering

Collaborative Filtering (CF) is applied in recommender systems to predict users’ preference through filtering the information or patterns. Privacy Preserving Collaborative Filtering (PPCF) aims to achieve privacy protection in the recommendation process, which has an increasing significance in recommender systems and thus attracted much interests in recent years. Existing PPCF methods are mainly based on cryptography, obfuscation, perturbation and differential privacy. They have high computational cost, low data quality and difficulties in calibrating the magnitude of noise. This paper proposes a (p, l, α)-diversity method that improves the existing k-anonymity method in PPCF, where p is attacker’s prior knowledge about users’ ratings and (l, α) is the diversity among users in each group to improve the level of privacy preserving. To achieve (l, α)-diversity, users in each equivalence class shall come from at least l (l < k) clusters in α clustering results. Therefore, we firstly apply Latent Factor Model (LFM) to reduce matrix sparsity. Then we propose an improved Maximum Distance to Average Vector (MDAV) microaggregation algorithm based on importance partitioning to increase the homogeneity among the records in each group which can retain better data quality in (p, l, α)-diversity model. Finally, we apply t-closeness in PPCF. Theoretical analysis and experimental results demonstrate our approach assures a higher level of privacy preserving and less information loss than existing methods.

Designing shilling attacks on disguised binary data

Privacy-preserving collaborative filtering methods are effectual ways of coping with information overload problem while protecting confidential data. Their success depends on the quality of the collected data. However, malicious entities might create fake profiles and insert them into user-item matrices of such filtering schemes. Hence, shilling attacks play an important role on the quality of data. Designing effective shilling attacks and performing robustness analysis of privacy-preserving collaborative filtering methods are receiving increasing attention. In this study, six shilling attack models are designed to attack binary disguised user-item matrices in privacy-preserving collaborative filtering methods. The attack models are applied to naive Bayesian classifier-based collaborating filtering scheme with privacy to measure its robustness against fake profiles. Empirical results on real data show that designing effective shilling attacks is still possible on perturbed binary ratings and na

Randomization-based Privacy-preserving Frameworks for Collaborative Filtering

Randomization-based privacy protection methods are widely used in collaborative filtering systems to achieve individual privacy. The basic idea behind randomization utilized in collaborative filtering schemes is to add randomness to original data in such a way so that required levels of accuracy and privacy can be achieved. Although there are various studies on privacy-preserving collaborative filtering using randomization, there are no well-defined privacy-preserving frameworks for collaborative filtering algorithms based on randomization. In this paper, we present eight randomization-based privacy-preserving frameworks for privacy protection in collaborative filtering schemes. We first group privacy-preserving methods into two broad categories. We then classify them based on private data. Final grouping is done while considering varying privacy concerns of individual users. The frameworks can be chosen according to individual users’ expectations and be utilized for privacy protection. The well-defined privacy-preserving frameworks form a basis for privacy protection based on randomized perturbation and randomized response techniques in collaborative filtering studies.

A k-anonymous approach to privacy preserving collaborative filtering

This article proposes a new technique for Privacy Preserving Collaborative Filtering (PPCF) based on microaggregation, which provides accurate recommendations estimated from perturbed data whilst guaranteeing user k-anonymity. The experimental results presented in this article show the effectiveness of the proposed technique in protecting users' privacy without compromising the quality of the recommendations. In this sense, the proposed approach perturbs data in a much more efficient way than other well-known methods such as Gaussian Noise Addition (GNA).

Privacy-Preserving Two-Party Collaborative Filtering on Overlapped Ratings

To promote recommendation services through prediction quality, some privacy-preserving collaborative filtering solutions are proposed to make e-commerce parties collaborate on partitioned data. It is almost probable that two parties hold ratings for the same users and items simultaneously; however, existing two-party privacy-preserving collaborative filtering solutions do not cover such overlaps. Since rating values and rated items are confidential, overlapping ratings make privacy-preservation more challenging. This study examines how to estimate predictions privately based on partitioned data with overlapped entries between two e-commerce companies. We consider both user-based and item-based collaborative filtering approaches and propose novel privacy-preserving collaborative filtering schemes in this sense. We also evaluate our schemes using real movie dataset, and the empirical outcomes show that the parties can promote collaborative services using our schemes.

Privacy-preserving Collaborative Filtering Using Randomized Response

Privacy-preserving Collaborative Filtering Using Randomized Response

Shilling Attacks against Privacy-Preserving Collaborative Filtering

Shilling Attacks against Privacy-Preserving Collaborative Filtering

Privacy-preserving Collaborative Filtering Using Randomized Response

This paper proposes a new privacy-preserving recommendation method classified into a randomized perturbation scheme in which a user adds random noise to the original rating value and a server provides a disguised data to allow users to predict rating value for unseen items. The proposed scheme performs perturbation in {\em randomized response} scheme, which preserves higher degree of privacy than that of additive perturbation. To address the accuracy reduction of the randomized response, the proposed scheme uses a {\em posterior probability distribution function}, derived from Bayes' estimation to reconstruction of the original distribution, to revise the similarity between items computed from the disguised matrix. A simple experiment shows the accuracy improvement of the proposed scheme.

A comparison of clustering-based privacy-preserving collaborative filtering schemes

Privacy-preserving collaborative filtering (PPCF) methods designate extremely beneficial filtering skills without deeply jeopardizing privacy. However, they mostly suffer from scalability, sparsity, and accuracy problems. First, applying privacy measures introduces additional costs making scalability worse. Second, due to randomness for preserving privacy, quality of predictions diminishes. Third, with increasing number of products, sparsity becomes an issue for both CF and PPCF schemes.In this study, we first propose a content-based profiling (CBP) of users to overcome sparsity issues while performing clustering because the very sparse nature of rating profiles sometimes do not allow strong discrimination. To cope with scalability and accuracy problems of PPCF schemes, we then show how to apply k-means clustering (KMC), fuzzy c-means method (FCM), and self-organizing map (SOM) clustering to CF schemes while preserving users’ confidentiality. After presenting an evaluation of clustering-based methods in terms of privacy and supplementary costs, we carry out real data-based experiments to compare the clustering algorithms within and against traditional CF and PPCF approaches in terms of accuracy. Our empirical outcomes demonstrate that FCM achieves the best low cost performance compared to other methods due to its approximation-based model. The results also show that our privacy-preserving methods are able to offer precise predictions.

P2P collaborative filtering with privacy

With the evolution of the Internet and e-commerce, collaborative filtering (CF) and privacy-preserving collaborative filtering (PPCF) have become popular. The goal in CF is to generate predictions with decent accuracy, efficiently. The main issue in PPCF, however, is achieving such a goal while preserving users' privacy. Many implementations of CF and PPCF techniques proposed so far are centralized. In centralized systems, data is collected and stored by a central server for CF purposes. Centralized storage poses several hazards to users because the central server controls users' data. In this work, we investigate how to produce naïve Bayesian classifier (NBC)-based recommendations while preserving users' privacy without using a central server. In a community of people, users might create a peer-to-peer (P2P) network. Through P2P network, users can communicate with each other and exchange data to produce predictions. We share the workload of prediction process and offer referrals efficiently using P2P network. We propose privacy-preserving schemes and analyze them in terms of accuracy, privacy, and efficiency. Our real data-based results show that our schemes offer accurate NBC-based predictions with privacy eliminating central server.