Problem In Recommender Systems Research Articles

KDCRec: Knowledge Distillation for Counterfactual Recommendation Via Uniform Data

The bias problems in recommender systems are an important challenge. In this paper, we focus on solving the bias problems via uniform data. Previous works have shown that simple modeling with a uniform data can alleviate the bias problems and improve the performance. However, the uniform data is usually few and expensive to collect in a real product. In order to use the valuable uniform data more effectively, we propose a novel and general knowledge distillation framework for counterfactual recommendation with four specific methods, including label-based distillation, feature-based distillation, sample-based distillation and model structure-based distillation. Moreover, we discuss the relation between the proposed framework and the previous works. We then conduct extensive experiments on both public and product datasets to verify the effectiveness of the proposed four methods. In addition, we explore and analyze the performance trends of the proposed methods on some key factors, and the changes in the distribution of the recommendation lists. Finally, we emphasize that counterfactual modeling with uniform data is a rich research area, and list some interesting and promising research topics worthy of further exploration. Note that the source codes are available at <uri>https://github.com/dgliu/TKDE_KDCRec</uri>.

Exploration in Sequential Recommender Systems via Graph Representations

Temporal graph networks are powerful tools for solving the cold-start problem in sequential recommender systems. However, graph models are still susceptible to feedback loops and data distribution shifts. The paper proposes a simple yet efficient graph-based exploration method for the mitigation of the aforementioned issues. It adopts the counter-based state exploration from reinforcement learning to the bipartite graph domain. We suggest a method that biases model predictions using Rooted PageRank towards locally unexplored items. The method shows competitive quality on the popular recommender systems benchmarks. We, also, provide an extensive qualitative analysis of experiment results and recommendations for the use of our method in production applications.

Directional Adversarial Training for Robust Ownership-Based Recommendation System

Machine learning algorithms are susceptible to cyberattacks, posing security problems in computer vision, speech recognition, and recommendation systems. So far, researchers have made great strides in adopting adversarial training as a defensive strategy. Single-step adversarial training methods have been proposed as viable solutions for improving model generality and resilience. However, there has been little study to address this issue in the context of ownership-based recommendations, which may fail to capture stable results. In this work, we adapt the single-step adversarial training for ownership recommendation systems. Our main technical contributions are as follows: (1) We propose Adversarial Consumption and Production Relationship (ACPR), a model that combines factorization machine and single-step adversarial training for ownership recommendations. It enables us to take advantage of modeling consumption-production interactions with a factorization machine instead of the conventional matrix factorization method for ownership recommendations. (2) We enrich the ACPR technique with directional adversarial training and call our technique Adversarial Consumption and Production Relationship-Aware Directional Adversarial Model (ACPR-ADAM). The idea behind our ACPR-ADAM is that instead of the worst perturbation direction, the perturbation direction in the embedding space is restricted to other examples in the current embedding space, allowing us to incorporate the collaborative signal into the training process. Lastly, through extensive evaluations on Reddit and Pinterest, we demonstrate that our proposed method outperforms state-of-the-art methods. Compared with CPR and ACPR on Reddit and Pinterest datasets, our proposed ACPR-ADAM achieves 93%, 88%, and 72%, 69% enhancement in terms of AUC and HR, respectively.

LBCF: A Link-Based Collaborative Filtering for Overfitting Problem in Recommender System

Recommender system (RS) suggests relevant objects to generate personalized service and minimize information overload issue. User-based collaborative filtering (UBCF) plays a dominant role in practical RSs. However, traditional UBCF suffers from a recommendation overfitting problem, i.e., recommendations generated by UBCF usually concentrate on popular items, resulting in lower diversity. In addition, UBCF cannot maintain a reasonable tradeoff between the accuracy and diversity of recommendations because raising the diversity is often accompanied by a decrease in accuracy. In this article, we propose a novel approach, namely link-based collaborative filtering, to enhance the recommendation accuracy and diversity simultaneously without employing additional complex information. First, a user–item bipartite network is constructed based on the user–item rating matrix of RSs. Then, a global–local weighted bipartite modularity is presented to conduct link partition so that links with the same community can not only be relatively denser but also own the same characteristic. Furthermore, redundant links are removed from each community by utilizing a link reduction algorithm so that neighborhood of a target user can be selected according to the more efficient nonredundant links. Finally, rating prediction is executed based on the rating information of neighborhood. Also, items owning the highest predicted rating scores will be recommended to the target user. Experimental results from three real datasets of RSs suggest that, without taking advantage of special additional data, our proposed approach outperforms the state-of-the-art studies and is able to generate personalized recommendations with satisfying accuracy and diversity simultaneously.

An enhanced framework for solving cold start problem in movie recommendation systems

&lt;p&gt;Recommendation systems (RSs) are used to obtain advice regarding decision-making. RSs have the shortcoming that a system cannot draw inferences for users or items regarding which it has not yet gathered sufficient information. This issue is known as the cold start issue. Aiming to alleviate the user’s cold start issue, the proposed recommendation algorithm combined tag data and logistic regression classification to predict the probability of the movies for a new user. First using alternating least square to extract product feature, and then diminish the feature vector by combining principal component analysis with logistic regression to predict the probability of genres of the movies. Finally, combining the most relevant tags based on similarity score with probability and find top N movies with high scores to the user. The proposed model is assessed using the root mean square error (RMSE), the mean absolute error (MAE), recall@N and precision@N and it is applied to 1M, 10M and 20M MovieLens datasets, resulting in an accuracy of 0.8806, 0.8791 and 0.8739.&lt;/p&gt;

Alleviating data sparsity problem in time-aware recommender systems using a reliable rating profile enrichment approach

Recommender systems use intelligent algorithms to learn a user’s preferences and provide them relevant suggestions. Lack of sufficient ratings – also known as data sparsity problem – often results in poor recommendation performance. The existing recommendation methods have mainly focused on designing recommenders with high accuracy without paying much attention to the reliability of the recommendations. On the other hand, the users’ preferences may vary over time and considering the time factor in the design process is crucial, which has been largely ignored in most of the existing recommenders. To deal with these issues, a novel recommendation method is proposed in this paper which incorporates temporal reliability and confidence measures into the recommendation process. First, the effectiveness of the users’ rating is measured using a probabilistic approach and ineffective rating profiles are enriched by adding some implicit ratings to them. The quality of the predictions is evaluated using a temporal reliability measure taking into account the changes of users’ preferences over time. Then, the ratings with low reliability values are recalculated using a novel procedure, which updates the target user’s neighborhood by removing ineffective users. This leads to a temporal confidence measure that is used to update the neighborhood to provide more reliable and accurate recommendations. The superiority of the proposed method over state-of-the-art recommendation methods is shown by conducting extensive experiments on three benchmark datasets.

A Novel Overlapping Method to Alleviate the Cold-Start Problem in Recommendation Systems

Recommendation systems (RSs) are tools for interacting with large and complex information spaces. They provide a personalized view of such spaces, prioritizing items likely to be of interest to the user. The main objective of RSs is to tool up users with desired items that meet their preferences. A major problem in RSs is called: “cold-start”; it is a potential problem called so in computer-based information systems which comprises a degree of automated data modeling. Particularly, it concerns the issue in which the system cannot draw any inferences nor have it yet gathered sufficient information about users or items. Since RSs performance is substantially limited by cold-start users and cold-start items problems; this research study takes the route for a major aim to attenuate users’ cold-start problem. Still in the process of researching, sundry studies have been conducted to tackle this issue by using clustering techniques to group users according to their social relations, their ratings or both. However, a clustering technique disregards a variety of users’ tastes. In this case, the researcher has adopted the overlapping technique as a tool to deal with the clustering technique’s defects. The advantage of the overlapping technique excels over others by allowing users to belong to multi-clusters at the same time according to their behavior in the social network and ratings feedback. On that account, a novel overlapping method is presented and applied. This latter is executed by using the partitioning around medoids (PAM) algorithm to implement the clustering, which is achieved by means of exploiting social relations and confidence values. After acquiring users’ clusters, the average distances are computed in each cluster. Thereafter, a content comparison is made regarding the distances between every user and the computed distances of the clusters. If the comparison result is less than or equal to the average distance of a cluster, a new user is added to this cluster. The singular value decomposition plus (SVD[Formula: see text]) method is then applied to every cluster to compute predictions values. The outcome is calculated by computing the average of mean absolute error (MAE) and root mean square error (RMSE) for every cluster. The model is tested by two real world datasets: Ciao and FilmTrust. Ultimately, findings have exhibited a great deal of insights on how the proposed model outperformed a number of the state-of-the-art studies in terms of prediction accuracy.

Rating Prediction for Mobile Applications via Collective Matrix Factorization Considering App Categories

The rapid development of Internet and mobile communication, smart phone, tablet computer and other mobile devices become widespread, they have become an indispensable part of life, changing the human lifestyle, way of working and learning style, it brings great convenience and fun to us. This paper mainly discusses the status of personalized recommendation system, mobile application and analyse the main problems of recommendation system. This research proposes a rising mobile application domain, which uses web crawling technique to get explicit feedback from a real-world mobile application market. The explicit feedback is the users’ ratings data on Apps and the corresponding category information. To establish a mobile application recommendation model based on user interest and categorization information and propose a novel collective matrix factorization algorithm. Through the crawled data, the improved algorithm, and the traditional collaborative. The traditional mobile application recommendation systems are mostly based on keyword search, popularity, download usage and categories, thus they do not make personalized recommendation to users, and invalid to find applications that users are interested in. Through the user’s historical behaviour, a personalized recommendation system can be established to recommend the applications that fit user’s interests effectively. This paper is based on the collective matrix factorization using user interest such as user’s ratings and category information, designs a personalized recommendation system for mobile applications by using the relationship between them.

Solving distribution problems in content-based recommendation system with gaussian mixture model

Recommendation systems play an important role in boosting purchasing consumption for many manufacturers by helping consumers find the most appropriate items. Furthermore, there are quite range of recommendation algorithms so far that can be efficient; however, a content-based algorithm is always the most popular, powerful, and productive method taken at the begin time of any project. In the negative aspect, somehow the accuracy of content-based algorithm results is still a concern that correlates to probabilistic similarity. In addition, the similarity calculation method is another crucial that affect the accuracy of content-based recommendation in probabilistic problems. In order to solve these problems, we propose a new content-based recommendation based on the Gaussian Mixture Model to improve the accuracy with more sensitive results for probabilistic recommendation problems. Our proposed method is experimented on a liquor dataset including six main flavour tastes, liquor main taste tags, and some other criteria. The method clusters n liquor records relied on n vectors of six dimensions into k group (k < n) before applying a formula to sort the results. Compared our proposed algorithm with three other popular models on the above dataset, the accuracy of the experimental results not only outweighs the comparison to those of three other models but also attain a very speedy response time in real-life applications.

Reinforcement Learning-based Product Delivery Frequency Control

Frequency control is an important problem in modern recommender systems. It dictates the delivery frequency of recommendations to maintain product quality and efficiency. For example, the frequency of delivering promotional notifications impacts daily metrics as well as the infrastructure resource consumption (e.g. CPU and memory usage). There remain open questions on what objective we should optimize to represent business values in the long term best, and how we should balance between daily metrics and resource consumption in a dynamically fluctuating environment. We propose a personalized methodology for the frequency control problem, which combines long-term value optimization using reinforcement learning (RL) with a robust volume control technique we termed "Effective Factor". We demonstrate statistically significant improvement in daily metrics and resource efficiency by our method in several notification applications at a scale of billions of users. To our best knowledge, our study represents the first deep RL application on the frequency control problem at such an industrial scale.

A General Offline Reinforcement Learning Framework for Interactive Recommendation

This paper studies the problem of learning interactive recommender systems from logged feedbacks without any exploration in online environments. We address the problem by proposing a general offline reinforcement learning framework for recommendation, which enables maximizing cumulative user rewards without online exploration. Specifically, we first introduce a probabilistic generative model for interactive recommendation, and then propose an effective inference algorithm for discrete and stochastic policy learning based on logged feedbacks. In order to perform offline learning more effectively, we propose five approaches to minimize the distribution mismatch between the logging policy and recommendation policy: support constraints, supervised regularization, policy constraints, dual constraints and reward extrapolation. We conduct extensive experiments on two public real-world datasets, demonstrating that the proposed methods can achieve superior performance over existing supervised learning and reinforcement learning methods for recommendation.

Active learning strategies for solving the cold user problem in model-based recommender systems

Customers of a webshop are often presented large assortments, which can lead to customers struggling finding their desired product(s), an issue known as choice overload. In order to overcome this issue, recommender systems are used in webshops to provide personalized product recommendations to customers. Though, model-based recommender systems are not able to provide recommendations to new customers (i.e., cold users). To facilitate recommendations to cold users we investigate multiple active learning strategies, and subsequently evaluate which active learning strategy is able to optimally elicit the preferences from the cold users in a matrix factorization context. Our model is empirically validated using a dataset from the webshop of de Bijenkorf, a Dutch department store. We find that the overall best-performing active learning strategy is PopError, an active learning strategy that measures the variance score for each item.

Canonical tensor scaling

In this paper we generalize the canonical positive scaling of rows and columns of a matrix to the scaling of selected-rank subtensors of an arbitrary tensor. We expect our results and framework will prove useful for sparse-tensor completion required for generalizations of the recommender system problem beyond a matrix of user-product ratings to multidimensional arrays involving coordinates based both on user attributes (e.g., age, gender, geographical location, etc.) and product/item attributes (e.g., price, size, weight, etc.).

Knowledge Graph enhanced Neural Collaborative Filtering with Residual Recurrent Network

Knowledge Graph (KG), which commonly consists of fruitful connected facts about items, presents an unprecedented opportunity to alleviate the sparsity problem in recommender system. However, existing KG based recommendation methods mainly rely on handcrafted meta-path features or simple triple-level entity embedding, which cannot automatically capture entities’ long-term relational dependencies for the recommendation. Specially, entity embedding learning is not properly designed to combine user-item interaction information with KG context information. In this paper, a two-channel neural interaction method named Knowledge Graph enhanced Neural Collaborative Filtering with Residual Recurrent Network (KGNCF-RRN) is proposed, which leverages both long-term relational dependencies KG context and user-item interaction for recommendation. (1) For the KG context interaction channel, we propose Residual Recurrent Network (RRN) to construct context-based path embedding, which incorporates residual learning into traditional recurrent neural networks (RNNs) to efficiently encode the long-term relational dependencies of KG. The self-attention network is then applied to the path embedding to capture the polysemy of various user interaction behaviours. (2) For the user-item interaction channel, the user and item embeddings are fed into a newly designed two-dimensional interaction map. (3) Finally, above the two-channel neural interaction matrix, we employ a convolutional neural network to learn complex correlations between user and item. Extensive experimental results on three benchmark datasets show that our proposed approach outperforms existing state-of-the-art approaches for knowledge graph based recommendation.

Knowledge Transfer via Pre-training for Recommendation: A Review and Prospect.

Recommender systems aim to provide item recommendations for users and are usually faced with data sparsity problems (e.g., cold start) in real-world scenarios. Recently pre-trained models have shown their effectiveness in knowledge transfer between domains and tasks, which can potentially alleviate the data sparsity problem in recommender systems. In this survey, we first provide a review of recommender systems with pre-training. In addition, we show the benefits of pre-training to recommender systems through experiments. Finally, we discuss several promising directions for future research of recommender systems with pre-training. The source code of our experiments will be available to facilitate future research.

From edge data to recommendation: A double attention-based deformable convolutional network

Recommender systems (RSs) have become crucial parts in most web-scale applications, and the data sparsity is still one of the serious problems in RSs. Recently, the data sparsity can be tremendously alleviated by making use of informative data obtained from applications of edge devices and deep learning technologies with powerful data processing ability. In the face of extremely sparse rating data, the rich semantic information in reviews and the powerful feature extraction ability of convolutional neural network (CNN) contribute greatly to the improvement of recommendation performance. However, due to the complexity of natural language semantic components, it is common for many phrases in reviews to be separated by other words. Therefore, for the semantic information with unfixed intervals, the fixed geometric structure of CNN may lead to insufficient understanding of the user intention. Moreover, the usefulness of different reviews and the importance of different words in each review are various, which is vital for accurate modeling. In this paper, we propose a Double Attention-based Deformable Convolutional Network called DADCN for recommendation. In the proposed DADCN, two parallel deformable convolutional networks, which adopt the word-level and review-level attention mechanisms, are designed to flexibly extract features of both users and items from reviews. The parallel deformable convolutional networks jointly learn user preferences and item attributes, which is helpful to deepen the understanding of users’ attitudes. The word-level and review-level attention mechanisms are applied to intensify the critical words and informative reviews by assigning relatively high attention weights to them. Extensive experimental results on four real-world datasets demonstrate that the proposed DADCN outperforms four baseline methods.

RBPR: A hybrid model for the new user cold start problem in recommender systems

The recommender systems aim to predict potential demands of users by analyzing their preferences and provide personalized recommendation services. User preferences can be inferred from explicit or implicit feedback data. Most existing collaborative filtering (CF) methods rely heavily on explicit feedback data. However, these methods perform poorly when rating data is sparse. In this paper, we deal with the extreme case of sparse data, i.e., the new user cold start problem. In order to overcome this problem, we propose a novel CF ranking model, which combines a rating-oriented approach of Probabilistic Matrix Factorization (PMF) and a pairwise ranking-oriented approach of Bayesian Personalized Ranking (BPR) together. Therefore, our proposed model makes full use of the explicit and implicit feedback data. Experiments on the constructed new user cold start datasets based on four public datasets demonstrate the effectiveness of the proposed model for cold start recommendation. Code for the proposed method is available in https://gitee.com/xia_zhaoqiang/recomender-systems-rbpr.

Research Problems in Recommender systems

With continuous growth of web applications around the globe, it is a challenge to find the suitable information needed for the user in a limited time.Number of handheld mobile devices is increasing and most of the business revolves around the correct search of the data. Without a proper recommender system it is very difficult to get required information from the web applications. Web applications use recommender systems to provide suitable data to users based on their choices and interests. For different kinds of needs different types of recommender systems have been proposed. Two most basic types of recommender systems are collaborative filtering recommender system and content based recommender system. Sometimes these two recommender systems are combined to increase the efficiency of a recommender system The generated new recommender system is known as hybrid recommender system.The purpose of this paper is to help readers understand the basics of recommender systems. This paper identifies key areas of research openly available for new researchers. After reading this paper new researchers can understand basic problems of recommender systems which need improvement and hence they can make those problems their area of research.

A Unified Framework for Cross-Domain and Cross-System Recommendations

Cross-Domain Recommendation (CDR) and Cross-System Recommendations (CSR) are two of the promising solutions to address the long-standing data sparsity problem in recommender systems. They leverage the relatively richer information, e.g., ratings, from the source domain or system to improve the recommendation accuracy in the target domain or system. Therefore, finding an accurate mapping of the latent factors across domains or systems is crucial to enhancing recommendation accuracy. However, this is a very challenging task because of the complex relationships between the latent factors of the source and target domains or systems. To this end, in this paper, we propose a Deep framework for both Cross-Domain and Cross-System Recommendations, called DCDCSR, based on Matrix Factorization (MF) models and a fully connected Deep Neural Network (DNN). Specifically, DCDCSR first employs the MF models to generate user and item latent factors and then employs the DNN to map the latent factors across domains or systems. More importantly, we take into account the rating sparsity degrees of individual users and items in different domains or systems and use them to guide the DNN training process for utilizing the rating data more effectively. Extensive experiments conducted on three real-world datasets demonstrate that DCDCSR framework outperforms the state-of-the-art CDR and CSR approaches in terms of recommendation accuracy.

Privacy-Preserving Matrix Factorization for Cross-Domain Recommendation

Cross-domain recommender systems are known to provide solutions to the cold start and data sparsity problems in recommender systems. This can be achieved by leveraging sufficient ratings and users' profiles in one domain to enhance accurate recommendations in another domain. However, domains with sufficient ratings are not willing to share their users' ratings with other recommender systems or domains due to users' privacy and legal concern. Hence this shows a need for a privacy-preserving mechanism that encourages secure knowledge transfer between different domains. This study proposes a privacy-preserving cross-domain recommender system based on matrix factorization. Specifically, the study formally described the privacy requirements of a cross-domain recommender system, which are different from a single domain recommender system. It designs a new framework for a privacy-preserving cross-domain recommender system and then utilized the somewhat homomorphic encryption (SWHE) scheme to ensure users' privacy. The SWHE scheme was used to encrypt users' ratings in different domains, shared latent factor approach was implemented between the domains and extracted knowledge was securely transferred from the source domain to the target domain. We prove that users' privacy is secured throughout the stages involved in the proposed protocol. Experiments on both synthetic and real datasets demonstrate the efficiency of our protocol.