Search

In response to the COVID-19 crisis, higher education institutions increasingly rely on e-learning systems. Indeed, the higher education market has become increasingly competitive with the addition of open education models. However, the abundance of accessible online courses makes it challenging to deliver education that meets student needs. Learners have diverse profiles based on their traits, motivations, prior knowledge, and learning preferences. Recently, much research has given attention to the importance of using the contextual parameters to perform more accurate recommendations. In this context, context-aware recommendation of pedagogical resources can deal with the issue of information overload, cold start problem and meeting the learner's preferences. This paper describes a context-aware recommender system that harness the learner's contextual information. Our proposed approach is called Context Embedding Deep Collaborative Filtering (CEDCF), which enriches the learner profile with extracted sentiments from previous interactions. The proposed approach comprises three models, called Generalized Matrix Factorzation (GMF) , Multilayer Perceptron (MLP) and Neural Matrix Factorization (NeuMF). The GMF and the MLP are respectively applied to the rating matrix and the contextual parameters. The outputs of these models are then fed into a neural network to perform rating prediction. To put our proposal into shape, we model a real-world application of a merged coursera dataset to recommend courses. The experimental evaluation shows relevant results attesting the efficiency of the proposed approach. [ABSTRACT FROM AUTHOR]

In today's digital era, the abundance of online services presents users with a daunting array of choices, spanning from streaming platforms to e-commerce websites, leading to decision fatigue. Recommendation algorithms play a pivotal role in aiding users in navigating this plethora of options, among which collaborative filtering (CF) stands out as a prevalent technique. However, CF encounters several challenges, including scalability issues, privacy implications, and the well-known cold start problem. This study endeavors to mitigate the cold start problem by harnessing the capabilities of natural language processing (NLP) applied to user-generated reviews. A unique methodology is introduced, integrating both supervised and unsupervised NLP approaches facilitated by sci-kit learn, utilizing benchmark datasets across diverse domains. This study offers scientific contributions through its novel approach, ensuring rigor, precision, scalability, and real-world relevance. It tackles the cold start problem in recommendation systems by combining natural language processing (NLP) with machine learning and collaborative filtering techniques, addressing data sparsity effectively. This study emphasizes reproducibility and accuracy while proposing an advanced solution that improves personalization in recommendation models. The proposed NLP-based strategy enhances the quality of user-generated content, consequently refining the accuracy of Collaborative Filtering-Based Recommender Systems (CFBRSs). The authors conducted experiments to test the performance of the proposed approach on benchmark datasets like MovieLens, Jester, Book-Crossing, Last.fm, Amazon Product Reviews, Yelp, Netflix Prize, Goodreads, IMDb (Internet movie Database) Data, CiteULike, Epinions, and Etsy to measure global accuracy, global loss, F-1 Score, and AUC (area under curve) values. Assessment through various techniques such as random forest, Naïve Bayes, and Logistic Regression on heterogeneous benchmark datasets indicates that random forest is the most effective method, achieving an accuracy rate exceeding 90%. Further, the proposed approach received a global accuracy above 95%, a global loss of 1.50%, an F-1 Score of 0.78, and an AUC value of 92%. Furthermore, the experiments conducted on distributed and global differential privacy (GDP) further optimize the system's efficacy. [ABSTRACT FROM AUTHOR]

E-commerce recommender systems (RS) nowadays are essential for promoting products. These systems are expected to offer personalized recommendations for users based on the user preference. This can be achieved by employing cutting-edge technology such as artificial intelligence (AI) and machine learning (ML). Tailored recommendations for users can boost user experience in using the application and hence increase income as well as the reputation of a company. The purpose of this study is to investigate popular ML methods for e-commerce recommendation and study the potential of ensemble methods to combine the strengths of individual approaches. These recommendations are derived from a multitude of factors, including users' prior purchases, browsing history, demographic information, and others. To forecast the interests and preferences of users, several techniques are chosen to be investigated in this study, which include Singular Value Decomposition (SVD), k-Nearest Neighbor Baseline (KNN Baseline) and CoClustering. In addition, several evaluation metrics including the fraction of concordant pairs (FCP), mean absolute error (MAE), root mean square error (RMSE) and normalized discounted cumulative gain (NDCG) will be used to assess how well different techniques work. To provide a better understanding, the outcomes produced in this study will be incorporated into a graphical user interface (GUI). [ABSTRACT FROM AUTHOR]

The recommendation system is a filtering system. It filters a collection of things based on the historical behavior of a user, it also tries to make predictions based on user preferences and make recommendations that interest customers. While incredibly useful, they can face various challenges affecting their performance and utility. Some common problems are, for example, when the number of users and items grows, the computational complexity of generating recommendations increases, which can increase the accuracy and precision of recommendations. So, for this purpose and to improve recommendation system results, we propose a recommendation system combining the demographic approach with collaborative filtering, our approach is based on users' demographic information such as gender, age, zip code, occupation, and historical ratings of the users. We cluster the users based on their demographic data using the k-means algorithm and then apply collaborative filtering to the specific user cluster for recommendations. The proposed approach improves the results of the collaborative filtering recommendation system in terms of precision and recommends diverse items to users. [ABSTRACT FROM AUTHOR]

Abstract The interaction data used for training recommender systems often exhibit a long-tail distribution. Such highly imbalanced data distribution results in an unfair learning process among items. Contrastive learning alleviates the above issue by data augmentation. However, it lacks consideration of the significant disparity in popularity between items and may even introduce false negatives during the data augmentation, misleading user preference prediction. To address this issue, we combine contrastive learning with a weighted model for negative validation. By penalizing identified false negatives during training, we limit their potential harm within the training process. Meanwhile, to tackle the scarcity of supervision signals for unpopular items, we design Popularity Associated Modeling to mine the correlation among items. Then we guide unpopular items to learn hidden features favored by specific users from their associated popular items, which provides effective supplementary information for their representation modeling. Extensive experiments on three real-world datasets demonstrate that our proposed model outperforms state-of-the-art baselines in recommendation performance, with Recall@20 improvements of 4.2%, 2.4% and 3.6% across the datasets, but also shows significant effectiveness in relieving popularity bias.

Traditional collaborative filtering recommendation algorithms suffer from low recommendation efficiency and poor accuracy when calculating similarities between users or items. To address this issue and improve the efficiency of recommendation systems, the paper introduces an algorithm called K-nearest neighbors and non-negative matrix factorization (KNNCNMF) collaborative filtering recommendation algorithm. When calculating the similarity between users or items, the algorithm extracts the latent factors of users and items through matrix decomposition, constructs a low-dimensional dense “user–item factor” matrix, and inputs it into the classifier for rating prediction, which replaces the complex similarity calculation and further improves the efficiency of the user–item similarity calculation. We use performance metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Precision, and Recall to measure our method. The experimental results show that compared to other algorithms, our method improves the MAE metric by 1.78% on average, the RMSE metric by 4.48% on average, the Precision metric by 4.66% on average, and the Recall metric by 7.95% on average. It proves the effectiveness of our proposed method. [ABSTRACT FROM AUTHOR]

The interaction data used for training recommender systems often exhibit a long-tail distribution. Such highly imbalanced data distribution results in an unfair learning process among items. Contrastive learning alleviates the above issue by data augmentation. However, it lacks consideration of the significant disparity in popularity between items and may even introduce false negatives during the data augmentation, misleading user preference prediction. To address this issue, we combine contrastive learning with a weighted model for negative validation. By penalizing identified false negatives during training, we limit their potential harm within the training process. Meanwhile, to tackle the scarcity of supervision signals for unpopular items, we design Popularity Associated Modeling to mine the correlation among items. Then we guide unpopular items to learn hidden features favored by specific users from their associated popular items, which provides effective supplementary information for their representation modeling. Extensive experiments on three real-world datasets demonstrate that our proposed model outperforms state-of-the-art baselines in recommendation performance, with Recall@20 improvements of 4.2%, 2.4% and 3.6% across the datasets, but also shows significant effectiveness in relieving popularity bias. [ABSTRACT FROM AUTHOR]

In light of the remarkable capacity of graph convolutional network (GCN) in representation learning, researchers have incorporated it into collaborative filtering recommendation systems to capture high-order collaborative signals. However, existing GCN-based collaborative filtering models still exhibit three deficiencies: the failure to consider differences between users' activity and preferences for items' popularity, the low-order feature information of users and items has been inadequately employed, and neglecting the correlated relationships among isomorphic nodes. To address these shortcomings, this paper proposes a degree-aware embedding-based multi-correlated graph convolutional collaborative filtering (Da-MCGCF). Firstly, Da-MCGCF combines users' activity and preferences for items' popularity to perform neighborhood aggregation in the user-item bipartite graph, thereby generating more precise representations of users and items. Secondly, Da-MCGCF employs a low-order feature fusion strategy to integrate low-order features into the process of mining high-order features, which enhances feature representation capabilities, and enables the exploration of deeper relationships. Furthermore, we construct two isomorphic graphs by employing an adaptive approach to explore correlated relationships at the isomorphic level between users and items. Subsequently, we aggregate the features of isomorphic users and items separately to complement their representations. Finally, we conducted extensive experiments on four public datasets, thereby validating the effectiveness of our proposed model. [ABSTRACT FROM AUTHOR]

Recommender Systems (RSs) have recently emerged as a practical solution to the information overload problem users face when searching for digital content. In general, RSs provide their respective users with specialized advice and guidance in order to make informed decisions on the selection of suitable digital content. This paper is a systematic literature review of recent (2011–2023) publications on RSs designed and developed in the context of education to support teachers in particular—one of the target groups least frequently addressed by existing RSs. A body of 61 journal papers is selected and analyzed to answer research questions focusing on experimental studies that include RS evaluation and report evaluation results. This review is expected to help teachers in better exploiting RS technology as well as new researchers/developers in this field in better designing and developing RSs for the benefit of teachers. An interesting result obtained through this study is that the recent employment of machine learning algorithms for the generation of recommendations has brought about significant RS quality and performance improvements in terms of recommendation accuracy, personalization and timeliness. [ABSTRACT FROM AUTHOR]

The study presents a Cluster-Based Cab Recommender System (CBCRS) designed to optimize cab services by suggesting the nearest locations with a higher likelihood of finding passengers. To achieve this, the system employs advanced clustering techniques to cluster historical cab pickup locations, identifying areas with higher passenger possibilities at specific times and days. The research aims to develop an algorithmic framework for CBCRS based on a hybrid clustering technique. The objectives of the study are twofold: first, to identify current clustering techniques used in clustering cab pickup geo-points, and second, to propose a framework for CBCRS based on the most efficient clustering technique. This framework will accept the current location of the cab driver and recommend the next nearest passenger pickup location. Additionally, the study compares and contrasts the proposed system with other clustering techniques using three standard datasets, evaluating them based on intrinsic measures such as the Calinski-Harabasz Index and Silhouette-Score. The paper concludes by evaluating and contrasting the proposed CBCRS framework with different clustering techniques, analyzing the results using statistical parameters. The findings reveal that the proposed CBCRS system generates better recommendations for the cab drivers using CBCRS hybrid clustering technique as compared to K-Means, BIRCH, DBSCAN clustering algorithms. [ABSTRACT FROM AUTHOR]

In recent years, graph neural networks (GNNs) have been widely applied in recommender systems. However, existing recommendation algorithms based on GNNs still face challenges in node aggregation and feature extraction processes because they often lack the ability to capture the interactions between users and items, as well as users' multiple intentions. This hinders accurate understanding of users' needs. To address the aforementioned issues, we propose a recommendation model called multi-intent-aware interactive graph convolutional network (Multi-IAIGCN). This model is capable of integrating multiple user intents and adopts an interactive convolution approach to better capture the information on the interaction between users and items. First, before the interaction between users and items begins, user intents are divided and mapped into a graph. Next, interactive convolutions are applied to the user and item trees. Finally, by aggregating different features of user intents, predictions of user preferences are made. Extensive experiments on three publicly available datasets demonstrate that Multi-IAIGCN outperforms existing state-of-the-art methods or can achieve results comparable to those of existing state-of-the-art methods in terms of recall and NDCG, thus verifying the effectiveness of Multi-IAIGCN. [ABSTRACT FROM AUTHOR]

Tourism is an important sector, serving as an avenue to show the natural resources of a country and inhabitants' hospitality. This sector creates several opportunities for building a potential market and enhancing economic activities where tourist spots and activities flourish. Despite the numerous benefits, tourism still requires significant improvement, particularly in the Philippines, where there are abundant beautiful places. Therefore, this study aimed to develop a recommender system based on users and content collaborative filtering to provide local and foreign tourists with viable information for experience improvement. The investigation focused on improving tourist satisfaction based on three aspects such as preferences, ratings, and reviews that add options for tourist spots, activities/itineraries, destinations, and others. The machine algorithm-based journey assistant (MAJA) was designed as an interface and agent in providing help to tourists. The mean average precision (MAP) and recall were used as evaluation metrics to better understand the ability of MAJA to offer personalized experiences to unique users. The results showed that integration of the system into tourism provided a smart platform for enhancing tourist experience and destination competitiveness. Consequently, successful implementation of the system is measured by two criteria, namely the degree of tourists' pleasure during trip and the capacity of MAJA to effectively transfer tourism to less popular and less "accessible" sector.

Abstract In recent years, graph neural networks (GNNs) have been widely applied in recommender systems. However, existing recommendation algorithms based on GNNs still face challenges in node aggregation and feature extraction processes because they often lack the ability to capture the interactions between users and items, as well as users’ multiple intentions. This hinders accurate understanding of users’ needs. To address the aforementioned issues, we propose a recommendation model called multi-intent-aware interactive graph convolutional network (Multi-IAIGCN). This model is capable of integrating multiple user intents and adopts an interactive convolution approach to better capture the information on the interaction between users and items. First, before the interaction between users and items begins, user intents are divided and mapped into a graph. Next, interactive convolutions are applied to the user and item trees. Finally, by aggregating different features of user intents, predictions of user preferences are made. Extensive experiments on three publicly available datasets demonstrate that Multi-IAIGCN outperforms existing state-of-the-art methods or can achieve results comparable to those of existing state-of-the-art methods in terms of recall and NDCG, thus verifying the effectiveness of Multi-IAIGCN.

Abstract 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.

One of the most effective and extensively used recommendation technique is collaborative filtering. Based on related users or items, collaborative filtering creates recommendations for the users. Similarity measures play a crucial role in Collaborative Filtering Recommender System. One useful similarity measure for a cold-start situation is the Proximity-Impact-Popularity (PIP) measure. The PIP measure is, nevertheless, subject to several limitations as it penalises users' multiple times throughout similarity computation in addition to ignoring their global rating behaviour. When a user rates the items, their overall rating behavior—whether they are lenient or strict—is referred to as global rating behaviour. In this study, we introduce an improved similarity metric to calculate similarity more accurately and generate high-quality recommendations. Our method takes into account both the user's rating behaviour and the percentage of co-rated items among users. Additionally, we have considered the computation complexity of the suggested work. In addition, to exhibit the performance of the proposed measure, empirical analysis has been done on real datasets. The results of the experiments performed on the dataset show that the suggested work takes precedence over the current similarity metrics. In comparison to state-of-the-art measurements, the suggested work exhibits improvements in MAE of 1.73%, RMSE of 4.01%, and F measure of 1.47% on an average. [ABSTRACT FROM AUTHOR]

The personalized product preference and decision-making recommendation systems are highly demanded to handle big data and to increase service quality of the e-commerce platforms in the competitive industries. Previous recommender systems were hard coded and only extracted items from the same category. With this tactic, customers are limited to viewing only one category of products; items with several categories cannot be viewed. A concern for the e-commerce sector, particularly in the healthcare and pharmaceutical industries, is the growth of consumer preferences, the issue of cold starts, and the huge number of stocks holding units for new items. Therefore, this study aims to develop a product recommendation system for an e-commerce platform which deals with health supplements. For this reason, collaborative and content-based filtering are combined to propose a hybrid recommender system. In the proposed hybrid model, user's actions are converted into implicit rating weightage first. Then, to tackle the problem of increasing customer preferences, collaborative filtering is used to generate user's rating for warm-start items. Moreover, content-based filtering is used to solve cold start problem by recommending products to the users based on the similarity of the products regardless of user profile. Term frequency- inverse document frequency (TF-IDF) algorithm is adopted to weight the feature from the dataset first, then it creates step-by-step cosine similarity table. Finally, the proposed hybrid model is evaluated based on error metrices, ranking metrices, and business metrics and then compared based on the standard benchmarking algorithms. The best algorithm is selected to be used for the system development. Finally, the proposed hybrid model is developed and integrated into the real online e-commerce platform for healthcare company to handle the large number of stocks keeping units, cold start issues, and increasing customer preferences. This study can assist the healthcare companies to recommend relevant products to their customers and to help them stay competitive in healthcare e-commerce industry. [ABSTRACT FROM AUTHOR]

Nowadays, Deep learning (DL) techniques have been proven successful as learning techniques in various research fields ranging from computer vision to social networks. The approach of DL is flourishing in the field of recommender systems (RS). Researchers have deployed metadata or auxiliary information using DL approaches in diverse applications in the last decade to achieve better recommendation accuracy. Thus, the metadata plays a vital role in obtaining a better user-item interaction. At the same time, existing techniques are based on fixed user and item factors. Therefore, the model does not correctly identify actual latent factors representation, resulting in a high prediction error. To handle this problem, a user metadata embedding using a deep autoencoder RS model called "Metadata Embedding Deep AutoEncoder (MEDAE)" based collaborative filtering is proposed. MEDAE model takes embeds user metadata such as demographics along with the rating data. The MEDAE model consists of an embedding layer, Encoder, and Decoder. The embedding layer generates embedding or latent features of the users, items, and metadata; Encoder receives concatenated features of the user, item, and metadata, then encodes the inputs and passes them to the decoder; and the decoder reconstructs the output. To test the effectiveness of proposed model Root Mean Squared Error and Mean Absolute Error measures are used. Different architectures (like Big-Small-Big (BSB) (5), BSB (3), Small-Big-Small (3), and SBS (5)) of the MEDAE model are evaluated on MovieLens datasets along with different parameters such as activation functions (ELU and SELU) and regularization and results concluded that the MEDAE with SBS (3) and ELU + SELU component improves 4% of RMSE and 2% MAE over the baseline methods. [ABSTRACT FROM AUTHOR]

Recommendation techniques play a vital role in recommending an actual product to an intended user. The recommendation also supports the user in the decision-making process. In recent years, collaborative filtering has been a widely used technique in recommender systems. A model-based CF technique called matrix factorization fills the user–item interaction matrix’s missing elements. One of the significant challenges in MF is filling those elements in a row or column. The user has a few observations about an item, leading to sparsity issues of collaborative filtering. Therefore, conventional MF alone is not suitable to address the new item or user problem. We propose an MF model that integrates user-side information to handle these issues. We integrate user-side information in terms of vectors and bias to overcome the sparsity problem of collaborative filtering. We exhaustively evaluate our model on real-world datasets for predicting the ratings. The experiment results and analysis demonstrate that the proposed approach improves predictions significantly compared to the state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

Recommender systems are generally used in several domains, like e-commerce sites and social networks. E-learning systems use recommendation techniques to facilitate and improve online learning. Educational platforms offer users the necessary pedagogical tools to create an enriched learning environment, fostering collaboration and resource sharing. Recommender System faces many challenges. Among issues: (1) cold-start in which new users and/or items having not prior information available in the system; (2) data sparsity where rated items number is very small contrary to unrated items; and (3) scalability where more training data is required. This study presents a recommender system that uses learner criteria, such as learner's past behavior, demographics information, performance data, collaborative filtering, and ratings to suggest pedagogical resources. The proposed system adopts a hybrid approach, combining two primary methods: popularity-based and collaborative filtering-based. This hybrid approach enhances a collaborative filtering approach with popularity to provide a starting point for new users. The popularity-based is specifically used to address the issue of cold-start for new users by providing primary recommendations. Additionally, we have used two collaborative filtering approaches. The SVD-based enhances the recommendation list for the new user and tackles the sparsity problem. Simultaneously, enhanced matrix factorization with deep neural network (DNN) outperforms traditional matrix factorization in terms of recommendation diversity and accuracy. Our system improves the accuracy and effectively responds to user needs. Our approach early findings show promising results. It scores for top-10 items a total recall of (0.47), a global precision of (0.20), and an accuracy of (0.87). [ABSTRACT FROM AUTHOR]

In recent times, recommender systems have acquired significant popularity as a solution to the issue of information overload. These systems offer personalised recommendations to users. The superiority of multi-criteria recommender systems over their single-criterion counterparts has been demonstrated, as the former are able to provide more precise recommendations by taking into account multiple dimensions of user preferences when rating items. The prevalent recommendation technique, matrix factorization of collaborative filtering, is hindered by the data sparsity problem of the user-item matrix. On the other hand, it is noteworthy that deep learning techniques have demonstrated significant potential in various research domains, including but not limited to image processing, pattern recognition, computer vision, and natural language processing. In recent times, there has been a surge in the utilisation of deep learning techniques in recommender systems, yielding promising outcomes. This study presents a novel approach to multi-criteria recommender systems through the utilisation of deep learning algorithms to mitigate the data sparsity issue. Specifically, deep autoencoders are utilised to uncover complex, non-linear, and latent relationships between users' multi-criteria preferences followed by matrix factorization technique, ultimately leading to more precise recommendations. The proposed model is evaluated by conducting the experiments on the multi-criteria dataset of Yahoo! Movies. According to the outcomes, the proposed approach outperforms the state of the art recommendation methods by generating more accurate and personalized recommendations. Also, it reduces the data sparsity up to 11% from the multi-criteria dataset. [ABSTRACT FROM AUTHOR]

Developing online educational platforms necessitates the incorporation of new intelligent procedures in order to improve long-term student experience. Presently, e-learning Recommender Systems rely on deep learning methods to recommend appropriate e-learning materials to the students based on their learner profiles. Fine-grained sentiment analysis (FSA) can be leveraged to enrich the recommender system. Users posted reviews and rating data are vital in accurately directing the student to the appropriate e-learning resources based on posted comments by comparable learners. Innovative has been made in this work to propose a new e-learning recommendation system based on individualization and FSA. A new framework is proposed based on collaborative filtering models (CFMs) integrating with fine-grained sentiment analysis (FSA) for hybrid recommendation (CFISAR) for effective recommendations. CFMs attempt to capture the learner's latent factors based on their selections of interest to build the learner profile. FSA models are introduced to deliver e-content with the highest ranked ratings related to the learner's area and interests based on the extracted learner model. Moreover, a new approach is proposed to update the system continuously and not keep it bound to certain items by adding new books, where the initial rating of these new books is predicted based on FSA models. CFISAR is explored utilizing six CFMs to generate the prediction matrix and derive the learner model, resulting in a low MSE of 0.699 for Asymmetric SVD. The system used multiplication word embeddings for stronger corpus representation that were trained on a dataset generated for an educational context, and leveraging the goodness of deep learning, which predicted an accuracy of 0.9264% for the Peephole algorithm, that performed better than other models. [ABSTRACT FROM AUTHOR]

Recommender systems are widely used in e-commerce, movies, music, social media, and other fields because of their personalized recommendation functions. The recommendation algorithm is used to capture user preferences, item characteristics, and the items that users are interested in are recommended to users. Matrix factorization is widely used in collaborative filtering algorithms because of its simplicity and efficiency. However, the simple dot-product method cannot establish a nonlinear relationship between user latent features and item latent features or make full use of their personalized information. The model of a neural network combined with an attention mechanism can effectively establish a nonlinear relationship between the potential features of users and items and improve the recommendation accuracy of the model. However, it is difficult for the general attention mechanism algorithm to solve the problem of attention interaction when the number of features between the users and items is not the same. To solve the above problems, this paper proposes an attention interaction matrix factorization (AIMF) model. The AIMF model adopts a symmetric structure using MLP calculation. This structure can simultaneously extract the nonlinear features of user latent features and item latent features, thus reducing the computation time of the model. In addition, an improved attention algorithm named slide-attention is included in the model. The algorithm uses the sliding query method to obtain the user's attention to the latent features of the item and solves the interaction problem among different dimensions of the user, and the latent features of the item. [ABSTRACT FROM AUTHOR]

In the current phase of technological development, Netflix has become a popular platform for entertainment. Often people feel overwhelmed when choosing a movie because of the variety of genres. To overcome this problem, a recommendation system is needed that can help people find the best movie according to their preferences. In addition, Twitter was used to collect tweets related to movies, which were then processed into rating values. The crawled dataset consisted of 855 movies and 44 user reviews (including data from IMDB, rotten tomatoes, and metacritic websites), for a total of 23,130 records. This research proposed to use the switching hybrid filtering (SHF) method combined with recurrent neural network (RNN) as classification. In SHF, the emphasis on rating prediction was initiated by the content based filtering method with RoBERTa, followed by switching to itembased collaborative filtering. This situation arose because the dataset had a sparseness of 74.46%. SHF provided accurate rating prediction with an MAE of 0.0617 and an RMSE of 0.1178. Nadam optimization with optimal learning rate in RNN classification gave the best results with an accuracy of 86.11%. The research successfully developed a method that proved effective and provided positive results, contributing to the development of a recommendation system designed to assist users in choosing movies based on their preferences. [ABSTRACT FROM AUTHOR]

Tourism is an important sector, serving as an avenue to show the natural resources of a country and inhabitants' hospitality. This sector creates several opportunities for building a potential market and enhancing economic activities where tourist spots and activities flourish. Despite the numerous benefits, tourism still requires significant improvement, particularly in the Philippines, where there are abundant beautiful places. Therefore, this study aimed to develop a recommender system based on users and content collaborative filtering to provide local and foreign tourists with viable information for experience improvement. The investigation focused on improving tourist satisfaction based on three aspects such as preferences, ratings, and reviews that add options for tourist spots, activities/itineraries, destinations, and others. The machine algorithm-based journey assistant (MAJA) was designed as an interface and agent in providing help to tourists. The mean average precision (MAP) and recall were used as evaluation metrics to better understand the ability of MAJA to offer personalized experiences to unique users. The results showed that integration of the system into tourism provided a smart platform for enhancing tourist experience and destination competitiveness. Consequently, successful implementation of the system is measured by two criteria, namely the degree of tourists' pleasure during trip and the capacity of MAJA to effectively transfer tourism to less popular and less "accessible" sector. [ABSTRACT FROM AUTHOR]

A recommender system (RecSys) is a smart solution that offers personalized items to users. Utilizing adequate information from historical transactions in the tourism industry such as items, users, ratings, and reviews becomes valuable input in providing personalized RecSys regarding smart decision-making for users in tourism. This study proposes a new approach for tourism RecSys development through a hybrid model combining User-Based Collaborative Filtering (UBCF), Demographic Filtering (DF), Aspect-Based Sentiment Analysis (ABSA), and Content-Boosted Collaborative Filtering (CBCF). This model utilizes six steps in the Cross-Industry Standard Process for Data Mining (CRISP-DM): business understanding, data understanding, data preparation, modeling, evaluation, and deployment. A tourism dataset has been produced by combining raw data from the TripAdvisor website with demographic information from Google Maps to enhance user profiles. The model performance was evaluated using the measurement of Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) through the 5 cycles of 10-Fold Cross Validation. Based on the average results of MAE and RMSE, CBCF has achieved performance improvements respectively at 84.7% and 82.3% compared to the performance of UBCF using 100% dense UI-Matrix. This study has succeeded in proposing a novelty model for tourism recommender systems in avoiding the cold-start problem and sparse matrix reduction through the synthetic data generation regarding performance improvement of the recommender system complemented with related aspects of recommended attractions.

The expansion of the e-commerce market has led to the challenge of information overload, necessitating the development of recommender systems. The recommender system aids users in decision-making by suggesting items that align with their preferences. However, conventional recommendation models rely solely on quantitative user behavior data, such as user ratings, and lead to limitations in recommendation performance due to the sparsity problem. To address these issues, recent research has leveraged convolutional neural networks (CNNs) to extract and incorporate semantic information from user reviews. However, several prior studies have a disadvantage in that they fail to account for the intricate interactions between users and items directly. In this study, we introduce a novel approach, the Review-based recommender system using Outer Product on CNN (ROP-CNN) model, which adeptly captures and incorporates semantic features from reviews to address the complex interactions between users and items using CNN. The experimental results, using real user-review datasets, demonstrate that the ROP-CNN model outperforms existing baseline models for prediction accuracy. And this study presents a novel theoretical and methodological perspective in recommendation research, suggesting a method that integrates user preference information from reviews into recommender systems by leveraging rich user-item interaction information.

Collaborative Filtering has achieved great success in capturing users’ preferences over items. However, existing techniques only consider limited collaborative signals, leading to unsatisfactory results when the user-item interactions are sparse. In this paper, we propose a Cross-grained Neural Collaborative Filtering model (CNCF), which enables recommendation more accurate and explainable. Specifically, we first construct four kinds of interaction graphs to model both fine-grained collaborative signals and coarse-grained collaborative signals, which can better compensate for the high sparsity of user-item interactions. Then we propose a fine-grained collaborative representation learning and design Light Attribute Prediction Networks ( $LAPN$ ) to capture the high-order attribute interactions and enhance the prediction accuracy. Finally, we propose a coarse-grained collaborative representation learning to represent user preferences based on diverse latent intent factors. The experiments demonstrate the high effectiveness of our proposed model.

Recommender systems become essential with the presence of the internet and social media. The perceived benefits of the recommender system can make it easier for users to find suitable products and recommend other products, specifically with lots of information. Recommender systems continue to develop over time. It has led many researchers to continue to find the latest approach and evaluation techniques by comparing the performance of previously existing recommender systems. The main approaches that are often used in recommender systems are Content-Based Filtering (CBF), Collaborative Filtering (CF), and Hybrid Filtering (HBF). This time, we focus on conducting a Systematic Literature Review (SLR) of several research articles and analyzing methods for algorithms developed in building recommender systems. The SLR method consists of three stages: planning, implementation, and reporting. The research used as a comparison is between 2019 and 2023 using various existing data sets. There were 72 primary studies, of which 46 employed the Collaborative Filtering approach, 11 used Content-Based filtering, and 15 used Hybrid Filtering. The results of this SLR process show the advantages and disadvantages of each method and type of evaluation developed in building a recommender system. Apart from that, several challenges arise with various existing problems. However, the model-based collaborative filtering method is one method that can minimize the problems of cold start, data sparsity, and scalability.

The cold start problem has always been a major challenge for recommender systems. It arises when the system lacks rating records for new users or items. Addressing the challenge of providing personalized recommendations in the cold start scenario is crucial. This research proposes a new hybrid recommender system named CoDFi-DL which combines demographic and enhanced collaborative filtering. The demographic filtering is performed through a deep neural network (DNN) and used to solve the new user cold start problem. The enhanced collaborative filtering component of our model focuses on delivering personalized recommendations through a neighborhood-based method. The major contribution in this research is the DNN-based demographic filtering which overcomes the new user cold start problem and enhances the collaborative filtering process. Our system significantly improves the relevancy of the recommendation task and thus provides personalized recommended items to cold users. To evaluate the effectiveness of our approach, we conducted experiments on real multi-label datasets, 1M and 100K MovieLens. CoDFi-DL recommender system showed higher performance in comparison with baseline methods, achieving lower RMSE rates of 0.5710 on the 1M MovieLens dataset and 0.6127 on the 100K MovieLens dataset. [ABSTRACT FROM AUTHOR]

The great progress in information and communication technology has led to a huge increase in data available. Traditional systems can't keep up with this growth and can't handle this huge amount of data. Recommendation systems are one of the most important areas of research right now because they help people make decisions and find what they want among all this data. This study looked at the research trends published in Google Scholar within the period 2018-2022 related to recommending, reviewing, analysing, and comparing ebooks research papers. At first, the research papers were collected and classified based on the recommendation model used, the year of publication, and then they were compared in terms of techniques, datasets utilised, problems, contributions, and evaluation methods used. It was found that many in-depth studies of book recommendation systems directly affect how those systems grow. Many researchers interested in book recommendation systems can learn about the many parts of the field by looking at how the study was put together. [ABSTRACT FROM AUTHOR]

To address user cold-start challenge in multimedia recommender systems, we proposed a new model named USBE in this paper. The model doesn't take the new user's personal and social information as the necessary parameters to solve cold-start challenge, and new user can complete cold-start by having a simple system experience. Based on the user-similarity and the discrimination of the multimedia items, the model can recommend suitable items for cold-start users and let users choose and give feedback independently. Our model is lightweight and low delay, and provides a new cold-start mode. To complement USBE model, we proposed a cyclic training multilayer perceptron model (Re-NN) to get the strategy of new user's user-similarity changes. Experiments on a real-world movie recommendation dataset Movielens show: Our model has good results and achieves state-of-the-art after 4 rounds of cold-start recommendations. [ABSTRACT FROM AUTHOR]

Graph convolution networks (GCNs) play an increasingly vital role in recommender systems, due to their remarkable relation modeling and representation capabilities. Concretely, they can capture high-order semantic correlations within sparse bipartite interaction graphs, thereby enhancing user–item collaborative encodings. Despite the exciting prospects, the existing GCN-based models mainly focus on user–item interactions and seldom consider effectiveness of the side item co-occurrence information on user behavior guidance, resulting in limited performance improvement. Therefore, we propose a novel side item co-occurrence information-aware GCN model. Specifically, we first decouple the original heterogeneous relation graph into corresponding user–item and item–item subgraphs for user–item interaction and item co-occurrence relation modeling. Thereafter, we conduct adaptive iterative aggregation on these subgraphs for user intention understanding and co-occurring item correlation perception. Finally, we present two semantic fusion strategies for sufficient user–item semantic collaborative learning, thereby boosting the overall recommendation performance. Extensive comparison experiments are conducted on three benchmark datasets to justify the superiority of our model. [ABSTRACT FROM AUTHOR]

A recommendation system (RS) can help overcome information overload issues by offering personalized predictions for users. Typically, RS considers the overall ratings of users on items to generate recommendations for them. However, users may consider several aspects when evaluating items. Hence, a multi-criteria RS considers n-aspects of items to generate more accurate recommendations than a single-criteria RS. This research paper proposes two deep encoder–decoder models based on shared learning for a multi-criteria RS, multi-modal deep encoder–decoder-based shared learning (MMEDSL) and multi-criteria deep encoder–decoder-based shared learning (MCEDSL). MMEDSL employs the shared learning technique by concentrating on the multi-modality concept in deep learning, while MCEDSL focuses on the training process to apply the shared learning technique. The shared learning captures useful shared information during the learning process since the multi-criteria may have hidden inter-relationships. A set of experiments were conducted to compare the proposed models with recent baseline approaches. The Yahoo! Movies multi-criteria dataset was utilized. The results demonstrate that the proposed models outperform other algorithms. In addition, the results show that integrating the shared learning technique with the RS produces precise recommendation predictions. [ABSTRACT FROM AUTHOR]

Recommender systems position themselves as powerful tools in the support of relevance and personalization, presenting remarkable potential in the area of marketing. The cold-start customer problematic presents a challenge within this topic, leading to the need of distinguishing user features and preferences based on a restricted set of transactional information. This paper proposes a hybrid recommender system that aims to leverage transactional and portfolio information as indicating characteristics of customer behaviour. Four independent systems are combined through a parallelised weighted hybrid design. The first individual system utilises the price, target age, and brand of each product to develop a content-based recommender system, identifying item similarities. Secondly, a keyword-based content system uses product titles and descriptions to identify related groups of items. The third system utilises transactional data, defining similarity between products based on purchasing patterns, categorised as a collaborative model. The fourth system distinguishes itself from the previous approaches by leveraging association rules, using transactional information to establish antecedent and precedence relationships between items through a market basket analysis. Two datasets were analysed: product portfolio and transactional datasets. The product portfolio had 17,118 unique products and the included 4,408,825 instances from 2 June 2021 until 2 June 2022. Although the collaborative system demonstrated the best evaluation metrics when comparing all systems individually, the hybridisation of the four systems surpassed each of the individual systems in performance, with a 8.9% hit rate, 6.6% portfolio coverage, and with closer targeting of customer preferences and smaller bias. [ABSTRACT FROM AUTHOR]