Collaborative Filtering Recommender Systems Research Articles

An Improved Fusion-Based Semantic Similarity Measure for Effective Collaborative Filtering Recommendations

Semantic-enhanced recommendation systems are promising approaches to overcome the sparsity and cold-start problems, which are hard to handle using the conventional collaborative filtering (CF) approaches. Further research is needed to effectively integrate ontologies into collaborative filtering recommender systems. This paper proposes an ontology-based semantic similarity measure to evaluate similarities between items and eventually generate accurate recommendations. The proposed semantic similarity measure termed fusion-based semantic similarity takes into account the semantics of ontological instances (i.e. items) inferred from a specific domain ontology, which is determined by analyzing the hierarchical relationships among the instances, as well as the features of the instances and their relationships to other instances. The new measure comprehensively captures the semantic knowledge associated with instances by exploiting all possible shared semantics between instances in a given domain ontology. Furthermore, this paper proposes a new semantic-enhanced hybrid recommendation approach as a result of combining the new semantic similarity measure with the standard item-based CF to enhance the quality of generated recommendations. In order to assess the effectiveness of our semantic-enhanced hybrid collaborative filtering method, a series of experiments were conducted to compare the performance of the proposed approach against well-established benchmark techniques. The reported experimental results consistently emphasize its superiority, demonstrating enhanced predictive abilities and a notable improvement in the quality of recommendations. More specifically, the proposed approach achieved notable 6% reduction in Mean Absolute Error (MAE) in certain cases, outperforming other benchmark techniques. Additionally, this study highlights the potential of using semantic-based similarity to enhance the performance of recommendation systems. Such enhancements address challenges within collaborative filtering, potentially leading to advancements in recommendation system design and optimization.

A highly scalable CF recommendation system using ontology and SVD-based incremental approach

In recent years, the need of recommender systems has increased to enhance user engagement, provide personalized services, and increase revenue, especially in the online shopping industry where vast amounts of customer data are generated. Collaborative filtering (CF) is the most widely used and effective approach for generating appropriate recommendations. However, the current CF approach has limitations in addressing common recommendation problems such as data inaccuracy recommendations, sparsity, scalability, and significant errors in prediction. To overcome these challenges, this study proposes a novel hybrid CF method for movie recommendations that combines the incremental singular value decomposition approach with an item-based ontological semantic filtering approach in two phases, online and offline. The ontology-based technique is leveraged to enhance the accuracy of predictions and recommendations. Evaluating our method on a real-world movie recommendation dataset using precision, F1 scores, and mean absolute error (MAE) demonstrates that our system generates accurate predictions while addressing sparsity and scalability issues in recommendation system. Additionally, our method has the advantage of reduced running time.

How Do Product Recommendations Help Consumers Search? Evidence from a Field Experiment

Product recommendations can benefit consumers’ online product search via multiple underlying mechanisms, such as showing products that offer them high value, facilitating navigation on the website, or exposing more product information. However, it is unclear ex ante which is the primary underlying mechanism that drives the benefits of product recommendations to consumers. We conducted a randomized field experiment to estimate the benefits of an item-based collaborative filtering (CF) recommendation system to consumers. We collect unique data on the affinity scores computed by an item-based CF algorithm to develop measures of a product’s net value and horizontal (taste) fit for consumers. Our results indicate that product recommendations help consumers search for higher-value products that are lower priced, fit their tastes better, or both. Besides that, we find that the ability to find higher-value products (rather than easy navigation or exposure to more product information) is the primary driver for consumers’ higher purchase probabilities under recommendations. We further find a higher benefit of recommendations in product categories with higher price dispersion and heterogeneity in consumers’ tastes, providing additional evidence for the lower price and better horizontal fit mechanisms. Finally, we find that when made available, consumers substitute their usage of other search tools on the website with product recommendations. Our findings have important implications for online retailers, policymakers, regulators, and item-based CF recommendation system design. This paper was accepted by D. J. Wu, information systems. Funding: This work was supported by the Public Utility Research Center of the University of Florida, the Hi! PARIS Fellowship, the HEC Foundation, and the Leavey School of Business at Santa Clara University [Grant 102720]. Supplemental Material: Data and the online appendix are available at https://doi.org/10.1287/mnsc.2023.4951 .

Differential privacy in collaborative filtering recommender systems: a review.

State-of-the-art recommender systems produce high-quality recommendations to support users in finding relevant content. However, through the utilization of users' data for generating recommendations, recommender systems threaten users' privacy. To alleviate this threat, often, differential privacy is used to protect users' data via adding random noise. This, however, leads to a substantial drop in recommendation quality. Therefore, several approaches aim to improve this trade-off between accuracy and user privacy. In this work, we first overview threats to user privacy in recommender systems, followed by a brief introduction to the differential privacy framework that can protect users' privacy. Subsequently, we review recommendation approaches that apply differential privacy, and we highlight research that improves the trade-off between recommendation quality and user privacy. Finally, we discuss open issues, e.g., considering the relation between privacy and fairness, and the users' different needs for privacy. With this review, we hope to provide other researchers an overview of the ways in which differential privacy has been applied to state-of-the-art collaborative filtering recommender systems.

Uninteresting Items: Concept and Its Application to Effective Collaborative Filtering in Recommender Systems

Recommender systems aim to predict user preferences by analyzing users' past behavior. Collaborative filtering (CF) is one of the key techniques employed in recommender systems that uses explicit ( e.g. , ratings) and implicit ( e.g. , browsing) feedback from users to predict unknown feedback, providing top- N recommendations. However, CF faces challenges when dealing with sparse data, which can decrease the accuracy of recommendations. To overcome these inherent challenges in recommender systems, this article introduces the concept of "uninteresting items" that have not been rated by a user, but are unlikely to be liked even when recommended. We then review our previous works that utilize both positive preferences from rated items and negative preferences from uninteresting items to improve recommendation accuracy. Specifically, we discuss a family of our eight CF methods that are assisted by the uninteresting items: Zero-injection (ZI), l -injection, Imputation, RAGAN, and Deep-ZI, which are designed for explicit feedback, as well as gOCCF, M-BPR, and CNS, which are designed for implicit feedback. Also, we report some evaluation results for showing their effectiveness.

Variational Bayesian inference for bipartite mixed-membership stochastic block model with applications to collaborative filtering

A network-based method applied to collaborative filtering in recommender systems is introduced in this paper. Specifically, a novel mixed-membership stochastic block model with a conjugate prior from the exponential family is proposed for bipartite networks. The analytical expression of the model is derived, and a variational Bayesian algorithm that is computationally feasible for approximating the untractable posterior distributions is presented. Extensive simulations show that the proposed model provides more accurate inference than competing methods with the presence of outliers. The proposed model is also applied to a MovieLens dataset for a real data application.

A Graph Neural Network Social Recommendation Algorithm Integrating the Multi-Head Attention Mechanism

Collaborative filtering recommendation systems are facing the data sparsity problem associated with interaction data, and social recommendations introduce user social information to alleviate this problem. Existing social recommendation methods cannot express the user interaction interest and social influence deeply, which limits the recommendation performance of the system. To address this problem, in this paper we propose a graph neural network social recommendation algorithm integrating multi-head attention mechanism. First, based on the user-item interaction graph and social network graph, the graph neural network is used to learn the high-order relationship between users and items and deeply extract the latent features of users and items. In the process of learning user embedding vector representation based on the social network graph, the multi-head attention mechanism is introduced to increase the importance of friends with high influence. Then, we make rating predictions for the target users according to the learned user embedding vector representation and item embedding vector. The experimental results on the Epinions dataset show that the proposed method outperforms the existing methods in terms of both Recall and Normalized Discounted Cumulative Gain.

Comparison of Real-Time and Batch Job Recommendations

Collaborative filtering recommendation systems are traditionally trained in a batch manner and are designed to produce personalized recommendations for a large number of users at the same time. However, in many industrial use cases, it is reasonable to produce recommendations in real time, taking account of very recent user interactions. In this work, we present the implementation of batch and real-time recommendation systems using the example of the RP3Beta model, a simple scalable graph-based model that outperforms multiple more advanced models. Our approach can be utilized by any recommendation system if user-to-item recommendations can be obtained based on item-to-item recommendations. We show that it covers multiple common recommendation models, especially collaborative filtering approaches where user features are not available. We also provide the results of A/B tests comparing these two approaches in a real-world scenario of a job recommendation task, conducted with almost 200,000 OLX users. We report at least 10% more users applying for recommended job ads when using a real-time instead of a batch approach. We believe that our results can help other organizations to take informed decisions about whether to make the effort of moving from a batch to a real-time recommendation setting.

Wisdom from the crowd: Can recommender systems predict employee turnover and its destinations?

AbstractCan algorithms that predict customer movie and shopping preferences also predict which employees are likely to leave and where they are likely to go, thus helping to retain talent? This study applies a type of machine learning (ML) technique, collaborative filtering (CF) recommender system algorithms, to investigate the comparison between satisfaction with the current job and potential satisfaction with job alternatives, which is inherent in theorizing about individual turnover decisions. The comparison of those anticipated ratings along with employee's current job satisfaction creates two operationalizations: the quantity of more desirable job alternatives and the quality (or extent of desirability) of job alternatives. To test the effectiveness of this novel approach, we applied recommender system algorithms to a longitudinal archival dataset of employees and had three main findings. First, the recommender system algorithms efficiently predicted job satisfaction based on just two sources of information (i.e., work history and job satisfaction in previous jobs), providing construct validity evidence for recommender systems. Second, both the quantity and the quality of more desirable job alternatives compared to the current job positively correlated with employees’ future turnover behavior. Finally, our CF recommender system algorithms predicted where employees moved to, and even more effectively if constraining the alternative jobs to the same occupation. We conclude with implications how recommender system algorithms can help scholars effectively test theoretical ideas and practitioners predict and reduce turnover.

Graph Convolution Network Recommender Rystem with Feature Embedding

Recommender system has become one of the most important parts of information services on the Internet today the model based on graph neural network has been proved to be a very effective method in collaborative filtering recommender system [1,15]. However, the recommender system based on graph network usually only uses the interactive relationship between users and items, and their feature information is often not used effectively. In this work, we combine the use of feature information with graph convolutional network to design a new recommendation algorithm, Graph Convolution Network recommender system with Feature Embedding, GCNFE. Tests on the Movie lens and Last-FM datasets show that the new model performs better than previous models in most cases.

Applying Evolutionary-based User Characteristic Clustering and Matrix Factorization to Collaborative Filtering for Recommender Systems

<p>In recent years, with the rise of numerous Internet service industries, recommender systems have been widely used as never before. Users can easily obtain the information, products or services they need from the Internet, and businesses can also increase additional revenue through the recommender system. However, in today’s recommender system, the data scale is very large, and the sparsity of the scoring data seriously affects the quality of the recommendation. Thus, this study intends to propose a recommendation algorithm based on evolutionary algorithm, which combines user characteristic clustering and matrix factorization. In addition, the exponential ranking selection technology is employed for evolutionary algorithm. The experiment result shows that the proposed algorithm can obtain better result in terms of four indicators, mean square error, precision, recall, and F score for two benchmark datasets.</p> <p> </p>

Resource Management and Optimization Method of Music Audio-Visual Archives under the Background of Big Data

With the development of music and video, music and video management has been in a relatively backward state. This study uses the logistic regression algorithm based on sigmoid functions to analyze and process the music audio-visual library of a music software and establishes the logistic regression dynamic model, which provides a scientific research method for this complex system. The LPC characteristic coefficient is extracted, and then the logistic regression model of music audio-visual archives resources is established. After completing the model, this study obtains the logistic regression model through a series of experiments, which effectively optimizes the management of music audio-visual archives. The specific experimental conclusions are as follows: through genre classification and emotion classification, users can search more music audio-visual archives resources. The comparison shows that the recommendation effect after filtering by the logistic regression model is better than that of the nonstandard collaborative filtering recommendation system. Finally, after practical applications, it is concluded that the model based on the logistic regression algorithm has made a good optimization conclusion for the resource management of music audio-visual archives.

E-Commerce Personalized Recommendation Based on Machine Learning Technology

As e-commerce offers more and more choices for users, its structure becomes more and more complicated. Inevitably, it brings about the problem of information overload. The solution to this problem is an e-commerce personalized recommendation system using machine learning technology. People often seem confused when facing extensive information and cannot grasp the key points. This paper studies the personalized recommendation technology of e-commerce: deeply analyzes the related technologies and algorithms of the e-commerce recommendation system and proposes the latest architecture of the e-commerce recommendation system according to the current development status of the e-commerce recommendation system. The system recommends accuracy and real-time requirements and divides the system into two parts: offline mining and online recommendation and analyzes and implements the functions and technologies of each part. User-based recommender systems, collaborative filtering recommender systems, and content-based recommender systems are analyzed, respectively. The personalized recommendation cannot only quickly help customers find the required commodity information in a wide range of complex information but also can compare more commodity information to help customers to judge. However, the existing recommendation system has some problems such as the lack of recommendation personality, the reduced relevance of recommendation, and the poor timeliness of recommendation. Finally, a recommendation system that combines three recommendation algorithms is designed, and experiments are carried out. The newly designed recommendation system is compared with three different recommendation systems, and a summary and outlook are made. Based on the introduction of the relevant theories, characteristics, and mainstream technologies of personalized recommendation based on machine learning, this document presents a constructive example of a model based on the factors that influence personalized e-commerce information recommendations in the retail sector. Through questionnaire surveys, we analyze and design the influencing factors for consumers to purchase personalized products after the survey and build a project using state-of-the-art field learning techniques. Through the model to test the eight hypotheses proposed in this paper, the results show that customer income level, customer online shopping experience, commodity prices, product quality, recommendation relevance, credit evaluation, and service quality will have a significant positive impact on shopping willingness and ultimately affect the customer’s shopping behavior. e-commerce platform can use this influencing factor to establish personalized information recommendation service mode.

Improved Collaborative Filtering Recommender System Based on Missing Values Imputation on E-Commerce

One of the important aspects in e-commerce is how to recommend a product to users accurately. To achieve this goal, many e-commerce starts to build and research about recommender system. Many methods can be used to build a recommender system, one of them is using the collaborative filtering technique. This technique often experiences data sparsity problem that can impact to the recommender system prediction accuracy. To solve this problem, we apply improved collaborative filtering. This method predicts the missing values in the user item rating matrix. First, we do an initial selection to determine potential users who have the same characteristics with the active user. After that, we calculate the average distance between the active user and the other selected user. Next, we calculate missing values prediction. Missing values predictions is only done for items that have never been rated by other’s selected user but has been rated by the active user. We used Amazon electronic product with high sparsity level in this research to simulate the actual condition of e-commerce. We used MAE and RMSE to measure prediction accuracy. The methods we apply succeeds to improve the prediction accuracy compare to the conventional collaborative filtering method. The average MAE for method that we apply is 0.78 and RMSE 1.07

Design of Garment Style Recommendation System Based on Interactive Genetic Algorithm.

Recommender systems provide users with product information and suggestions, which has gradually become an important research tool in e-commerce IT technology, which has attracted a lot of attention of researchers. Collaborative filtering recommendation technology has been the most successful recommendation technology so far, but there are two major problems—recommendation quality and scalability. At present, research at home and abroad mainly focuses on recommendation quality, and there is less discussion on scalability. The scalability problem is that as the size of the system increases, the response time of the system increases to a point where users cannot afford it. Existing solutions often result in a significant drop in recommendation quality while reducing recommendation response time. In this paper, the clustering analysis subsystem based on the genetic algorithm is innovatively introduced into the traditional collaborative filtering recommendation system, and its design and implementation are given. In addition, when obtaining the nearest neighbors, only the clustered users of the target user are searched, making it a collaborative filtering recommender system based on genetic clustering. The experimental results show that the response time of the traditional collaborative filtering recommender system increases linearly with the increase in the number of users while the response time of the collaborative filtering recommender system based on genetic clustering remains unchanged with the increase in the number of users. On the other hand, the recommendation quality of the collaborative filtering recommender system based on genetic clustering is basically not degraded compared with that of the traditional collaborative filtering recommender system. Therefore, the collaborative filtering recommender system based on genetic clustering can effectively solve the scalability problem of the collaborative filtering recommender system.

Feature-blind fairness in collaborative filtering recommender systems

Recommender systems were originally proposed for suggesting potentially relevant items to users, with the unique objective of providing accurate suggestions. These recommenders started being adopted in several domains, and were identified as generating biased results that could harm the data items being recommended. The exposure in generated rankings, for instance in a job candidate selection situation, is supposed to be fairly distributed among candidates, regardless of their sensitive attributes (gender, race, nationality, age) for promoting equal opportunities. It can happen, however, that no such sensitive information is available in the data applied for training the recommender, and in this case, there is still space for biases that can lead to unfair treatment, named Feature-Blind unfairness. In this work, we adopt Variational Autoencoders (VAE), considered as the state-of-the-art technique for Collaborative Filtering (CF) recommendations, and we present a framework for addressing fairness when having only access to information about user-item interactions. More specifically, we are interested in Position and Popularity Bias. VAE loss function combines two terms associated with accuracy and quality of representation; we introduce a new term for encouraging fairness, and demonstrate the effect of promoting fair results despite of a tolerable decrease in recommendation quality. In our best scenario, position bias is reduced by 42% despite a reduction of 26% in recall in the top 100 recommendation results, compared to the same situation without any fairness constraints.

Optimization of fuzzy similarity by genetic algorithm in user‐based collaborative filtering recommender systems

AbstractThe most important subjects in the memory‐based collaborative filtering recommender system (RS) are to accurately calculate the similarities between users and finally finding interesting recommendations for active users. The main purpose of this research is to provide a list of the best items for recommending in less time. The fuzzy‐genetic collaborative filtering (FGCF) approach recommends items by optimizing fuzzy similarities in the continuous genetic algorithm (CGA). In this method, first, the crisp values of user ratings are converted to fuzzy ratings, and then the fuzzy similarities are calculated. Similarity values are placed into the genes of the genetic algorithm, optimized, and finally, they are used in fuzzy prediction. Therefore, the fuzzy system is used twice in this process. Experimental results on RecSys, Movielens 100 K, and Movielens 1 M datasets show that FGCF improves the collaborative filtering RS performance in terms of quality and accuracy of recommendations, time and space complexities. The FGCF method is robust against the sparsity of data due to the correct choice of neighbours and avoids the users' different rating scales problem but it not able to solve the cold‐start challenge.

Three-way Naive Bayesian collaborative filtering recommendation model for smart city

Smart city construction has penetrated all aspects of human life. People can continue to feel the convenience of life brought by new technologies and new services. In order to further optimize the smart city system, aiming at the sparsity of scoring data and the difficulty of two-way decisions to deal with uncertain decisions in the recommendation algorithm, a three-way Naive Bayesian Collaborative Filtering Recommendation model (3NBCFR) for smart city is constructed by integrating Naive Bayesian, three-way decisions and collaborative filtering algorithm. Firstly, we consider the influence of item attributes on user scoring. Naive Bayesian classifier is used to score unrated items, which can effectively fill in the missing values in the score matrix and solve the problem of data sparsity. Secondly, to ensure sustainable recommendations for uncertain goals, the three-way decisions is introduced into the collaborative filtering recommendation system, three-way recommendation rules are formulated, and 3NBCFR is constructed. Finally, the model is applied to the movie recommendation system. The experiment on Movielens shows that compared with the traditional collaborative filtering recommendation algorithm, 3NBCFR algorithm reduces the recommendation cost and improves the recommendation quality. At the same time, it lays a foundation for promoting the construction of smart city.

CD-SemMF: Cross-Domain Semantic Relatedness Based Matrix Factorization Model Enabled With Linked Open Data for User Cold Start Issue

Personalized recommendations to cold start user is one of the significant challenges in information filtering systems. Most of the existing systems inherited the idea of collaborative filtering (CF) and avoids the item metadata. For instance, consider a book domain whose metadata are author, publisher, language...etc. The elimination of item metadata is due to the diverse nature of item attributes in a cross-domain environment. Because of this ideology, it is hard for the filtering systems to provide better recommendations to cold start users. Cold start users are people who are new to a domain whose preferences are unknown. A novel model is proposed in this research work called Cross-Domain Semantic Relatedness based Matrix Factorization model (CD-SemMF) resolves the user cold start issue in collaborative filtering recommender system by exploiting Linked Open Data (LOD). “DBpedia” is a widely used LOD resource that contains semantic information about different domains, which is used in this research work to resolve the above-said problem. Here, the metadata available in LOD connects the items preferred by the target user from various domains. The fundamental knowledge graph links items from various domains and also benefits from cross-domain information. “LOD Semantic-Relatedness Measure” a new measure is proposed which calculates the closeness of items across domains. Semantic relatedness is measured instead of similarity in this research because the cross-domain item attributes are diverse. The Alternating Least Square method is applied here to learn the user preferences. The proposed model provides relevant, personalized recommendations for the target new user with the user preferences gained from the source domain and by exploiting item semantic relatedness. Experimental evaluation is done on Facebook and Amazon datasets. It is observed from the result that the proposed CD-SemMF model gives better recommendations in target domain for new users than the baseline methods.

SVD- Initialized K-Means Clustering for Collaborative Filtering Recommender Systems

SVD- Initialized K-Means Clustering for Collaborative Filtering Recommender Systems