Meta Matrix Factorization for Federated Rating Predictions

The rapid growth of edge data generated by mobile devices and applications deployed at the edge of the network has exacerbated the problem of information overload. As an effective way to alleviate information overload, recommender system can improve the quality of various services by adding application data generated by users on edge devices, such as visual and textual information, on the basis of sparse rating data. The visual information in the movie trailer is a significant part of the movie recommender system. However, due to the complexity of visual information extraction, data sparsity cannot be remarkably alleviated by merely using the rough visual features to improve the rating prediction accuracy. Fortunately, the convolutional neural network can be used to extract the visual features precisely. Therefore, the end-to-end neural image caption (NIC) model can be utilized to obtain the textual information describing the visual features of movie trailers. This paper proposes a trailer inception probabilistic matrix factorization model called Ti-PMF, which combines NIC, recurrent convolutional neural network, and probabilistic matrix factorization models as the rating prediction model. We implement the proposed Ti-PMF model with extensive experiments on three real-world datasets to validate its effectiveness. The experimental results illustrate that the proposed Ti-PMF outperforms the existing ones.

Recommender system is an effective way to help users to obtain the personalized and useful information. However, due to complexity and dynamic, the traditional recommender system cannot work well in mobile environment. In this paper, we propose a restaurant recommender system in mobile environment. This recommender system adopts a user preference model by using the features of user's visited restaurants, and also utilizes the location information of user and restaurants to dynamically generate the recommendation results. Baidu map cloud service is used to implement the proposed recommender system. The result of a case study shows that the proposed restaurant recommender system can effectively utilize user's preference and the location information to recommend the personalized and suitable restaurants for different users.

In order to tackle the problem of information overload and effective recommendation based on users' preference, need, and interest a number of research contributions has been made for the development of recommender systems. However, certain challenges, such as data sparsity, profiling attack, and black-box recommendation still exist and hamper their prediction accuracy. In this paper, we propose a hybrid approach to predict user ratings by incorporating both trust and context of the users in traditional recommender systems using collaborative filtering method. The similarity between two users is computed using both trust value and context-based similarity. The trust value is based on three trust statements -- rating deviation, emotions, and reviews helpfulness. On the other hand, context-based similarity is based on four contextual features -- companion, place, day, and priority. The performance of the proposed trust- and context-based hybrid approach is analyzed using mean absolute error and root mean square error on a real dataset generated from two movie data sources (IMDB and Rotten Tomatoes), and it performs significantly better in comparison to some of the standard baseline methods. The rating prediction using only trust statements gives better results in comparison to other collaborative filtering approaches, such as user-based and item-based filtering approaches. Similarly, context-based collaborative filtering approach also outperforms standard collaborative filtering approaches. In addition, rating prediction using both trust- and context-based features performs better in comparison to only trust-based or context-based approaches.

Social trust model for rating prediction in recommender systems: Effects of similarity, centrality, and social ties

Recommender systems have been an efficient strategy to deal with information overload by producing personalized predictions. Recommendation systems based on deep learning have accomplished magnificent results, but most of these systems are traditional recommender systems that use a single rating. In this work, we introduce a multi-criteria collaborative filtering recommender by combining deep neural network and matrix factorization. Our model consists of two parts: the first part uses a fused model of deep neural network and matrix factorization to predict the criteria ratings and the second one employs a deep neural network to predict the overall rating. The experimental results on two datasets, including a real-world dataset, show that the proposed model outperformed several state-of-the-art methods across different datasets and performance evaluation metrics.

Recommender systems are essential engines to deliver product recommendations for e-commerce businesses. Successful adoption of recommender systems could significantly influence the growth of marketing targets. Collaborative filtering is a type of recommender system model that uses customers' activities in the past, such as ratings. Unfortunately, the number of ratings collected from customers is sparse, amounting to less than 4%. The latent factor model is a kind of collaborative filtering that involves matrix factorization to generate rating predictions. However, using only matrix factorization would result in an inaccurate recommendation. Several models include product review documents to increase the effectiveness of their rating prediction. Most of them use methods such as TF-IDF and LDA to interpret product review documents. However, traditional models such as LDA and TF-IDF face some shortcomings, in that they show a less contextual understanding of the document. This research integrated matrix factorization and novel models to interpret and understand product review documents using LSTM and word embedding. According to the experiment report, this model significantly outperformed the traditional latent factor model by more than 16% on an average and achieved 1% on an average based on RMSE evaluation metrics, compared to the previous best performance. Contextual insight of the product review document is an important aspect to improve performance in a sparse rating matrix. In the future work, generating contextual insight using bidirectional word sequential is required to increase the performance of e-commerce recommender systems with sparse data issues.

In many recommender systems, users express item opinions through two kinds of behaviors: giving preferences and writing detailed reviews. As both kinds of behaviors reflect users' assessment of items, review enhanced recommender systems leverage these two kinds of user behaviors to boost recommendation performance. On the one hand, researchers proposed to better model the user and item embeddings with additional review information for enhancing preference prediction accuracy. On the other hand, some recent works focused on automatically generating item reviews for recommendation explanations with related user and item embeddings. We argue that, while the task of preference prediction with the accuracy goal is well recognized in the community, the task of generating reviews for explainable recommendation is also important to gain user trust and increase conversion rate. Some preliminary attempts have considered jointly modeling these two tasks, with the user and item embeddings are shared. These studies empirically showed that these two tasks are correlated, and jointly modeling them would benefit the performance of both tasks.

Knowledge graphs (KGs) provide rich semantic and structured knowledge to address data sparsity in traditional recommendation systems, yet efficiently leveraging KGs for better performance remains challenging. This paper proposes KIDRec, a knowledge graph recommendation framework integrating knowledge interaction and diffusion-augmented learning. This framework captures deep KG semantics via high-order entity interactions and uses a diffusion model to generate robust knowledge representations through “noise injection and denoising” on collaborative user embeddings and item embeddings. It then incorporates an adaptive contrastive learning module with hard negative sample enhancement and a dynamic temperature coefficient. Experiments on Book-Crossing, Last.FM, and MovieLens20m show KIDRec outperforms baselines in AUC (0.782, 0.877, 0.979) and F1 (0.682, 0.796, 0.932), with relative AUC improvements of 0.031, 0.032, and 0.002, respectively, while alleviating data sparsity.

We propose a new sentiment information-based attention mechanism that helps to identify user reviews that are more likely to enhance the accuracy of a rating prediction model. We hypothesis that highly polarised reviews (strongly positive or negative) are better indicators of the users’ preferences and that this sentiment polarity information helps to identify the usefulness of reviews. Hence, we introduce a novel neural network rating prediction model, called SentiAttn, which includes both the proposed sentiment attention mechanism as well as a global attention mechanism that captures the importance of different parts of the reviews. We show how the concatenation of the positive and negative users’ and items’ reviews as input to SentiAttn, results in different architectures with various channels. We investigate if we can improve the performance of SentiAttn by fine-tuning different channel setups. We examine the performance of SentiAttn on two well-known datasets from Yelp and Amazon. Our results show that SentiAttn significantly outperforms a classical approach and four state-of-the-art rating prediction models. Moreover, we show the advantages of using the sentiment attention mechanism in the rating prediction task and its effectiveness in addressing the cold-start problem.

The success of applying deep learning to many domains has gained strong interest in developing new revolutionary recommender systems. However, there are little works studying these systems that employ deep learning; additionally, there is no study showing how to combine the users and items embedding with deep learning to enhance the effectiveness of the recommender systems. Therefore, this paper proposes a novel approach called RecDNNing with a combination of embedded users and items combined with deep neural network. The proposed recommendation approach consists of two phases. In the first phase, we create a dens numeric representation for each user and item, called user embedding and item embedding, respectively. Following that, the items and users embedding are averaged and then concatenated before being fed into the deep neural network. In the second phase, we use the model of the deep neural network to take the concatenated users and items embedding as the inputs in order to predict the scores of rating by applying the forward propagation algorithm. The experimental results on MovieLens show that the proposed RecDNNing outperforms state-of-the-art algorithms.

Context-aware recommender systems in mobile environment: On the road of future research

With the rapid growth of information in the Internet era, people are in great need of recommendation methods to filter information. At present, recommendation methods which based on heterogeneous information network (HIN) have attracted wide attention. Recently, HIN-based recommendation methods need to be modeled from two aspects: node structural association and semantic association. To this end, we propose a graph contrastive learning model based on structural and semantic view for HIN recommendation (GCL-SS). GCL-SS utilizes U-I interactive view to obtain node structural embeddings, and utilizes U-I semantic view to obtain node semantic embeddings. Based on these two kinds of embeddings, we establish a self-supervised contrastive learning mechanism to effectively integrate structural information and semantic information of user (item) nodes in HIN, and finally learn a more discriminative user (item) embedding. In addition, in order to strengthen the semantic association between nodes, we innovatively utilize time sequence encoder (TSE), such as LSTM, to encode semantic homogeneous network decomposed by HIN in U-I semantic view. At last, based on the user and item embeddings, we adopt bilinear decoder to model the potential association between user and item, so as to realize rating prediction of user to item. The experimental results on three real datasets confirm that our GCL-SS model performs better than state-of-the-art recommendation methods in rating prediction task. In addition, the results of four ablation experiments indicate that our GCL-SS model can effectively improve the performance of rating prediction in recommendation.

In order to overcome the problems of poor operation function and low recommendation accuracy of the traditional high-quality curriculum resource recommendation system, this paper proposes and designs a new high-quality Curriculum Resource Recommendation System for educational technology specialty. With the support of the system framework, the system hardware module is designed, mainly including user registration and login module and high-quality course resource recommendation module, so as to realize the recommendation function of the system. The system software part constructs the user’s high-quality curriculum resources mining classification tree according to the user’s preference attribute value of high-quality curriculum resources, and obtains the recommended basic data. On this basis, the user’s preference for course resources is calculated, so as to complete the recommendation of high-quality course resources. The experimental results show that compared with the traditional recommendation system, the functions of the designed system have passed the test and have high recommendation accuracy.

Cross-domain recommendation exploits the rich data from source domain to solve the cold-start problem of target domain. Considering the recommendation system contains some user private information, how to provide accurate suggestions for cold-start users on the basis of protecting privacy is an important issue. Federated recommendation systems keep user private data on mobile devices to protect user privacy. However, compared to federated single-domain recommendation, federated cross-domain recommendation needs to train more models, making resource-constrained mobile devices infeasible to run large-scale models. In view of this, we design a meta embedding mapping method for federated cross-domain recommendation called MetaEM. The training stage of MetaEM includes pretraining and mapping. The pretrain stage learns user and item embeddings of source domain and target domain respectively. Items embeddings are divided into common and private. The common embeddings are shared by all users, and we train a meta-network to generate private embeddings for each user. The mapping stage learns to transfer user embeddings from source domain to target domain. In order to alleviate the negative impact of users with low number of ratings on mapping model, we employ a task-oriented optimization method. We implement the MetaEM prototype on large real-world datasets and extensive experiments demonstrate that MetaEM achieves the best performance and is more compatible with complicated models compared to other state-of-the-art baselines.KeywordsMeta learningCross-domain recommendationFederated learningCold-startEmbedding mapping

Location-based social networks (LBSNs), such as Gowalla, Facebook, Foursquare, Brightkite, and so on, have attracted millions of users to share their social friendship and their locations via check-ins in the past few years. Plenty of valuable information is accumulated based on the check-in behaviors, which makes it possible to learn users’ moving patterns as well as their preferences. In LBSNs, point-of-interest (POI) recommendation is one of the most significant tasks because it can help targeted users explore their surroundings as well as help third-party developers provide personalized services. Matrix factorization is a promising method for this task because it can capture users’ preferences to locations and is widely adopted in traditional recommender systems such as movie recommendation. However, the sparsity of the check-in data makes it difficult to capture users’ preferences accurately. Geographical influence can help alleviate this problem and have a large impact on the final recommendation result. By studying users’ moving patterns, we find that users tend to check in around several centers and different users have different numbers of centers. Based on this, we propose a Multi-center Gaussian Model (MGM) to capture this pattern via modeling the probability of a user’s check-in on a location. Moreover, users are usually more interested in the top 20 or even top 10 recommended POIs, which makes personalized ranking important in this task. From previous work, directly optimizing for pairwise ranking like Bayesian Personalized Ranking (BPR) achieves better performance in the top- k recommendation than directly using matrix matrix factorization that aims to minimize the point-wise rating error. To consider users’ preferences, geographical influence and personalized ranking, we propose a unified POI recommendation framework, which unifies all of them together. Specifically, we first fuse MGM with matrix factorization methods and further with BPR using two different approaches. We conduct experiments on Gowalla and Foursquare datasets, which are two large-scale real-world LBSN datasets publicly available online. The results on both datasets show that our unified POI recommendation framework can produce better performance.