Search

The rapid increase in fake reviews in e-commerce presents a considerable challenge, as it misleads consumers and compromises market integrity. Recently, graph-based models for detecting fake reviews have emerged as promising solutions. However, conventional methods and basic graph neural network techniques often fail to recognize the deceptive tactics of fraudsters, negatively affecting their efficacy. In response, we propose the label-consistent and hierarchical-relation-aware graph contrastive learning (LRGCL) framework, which is specifically designed to identify fake reviews, even when fraudsters employ various disguises. This framework implements a label-consistent graph contrastive learning approach to distinctly identify positive and negative samples, ensuring clear separation between authentic and fraudulent nodes despite feature camouflage. Additionally, LRGCL classifies negative samples into several categories using a hierarchical approach that leverages relationships within a multi-relation graph, thereby counteracting relation camouflage. The integration of these two strategies significantly reduces the effectiveness of these camouflages, leading to more precise detection of fake reviews. Empirical tests using real-world datasets validate the superior efficacy of LRGCL, as it surpasses current leading methods across key performance metrics. [ABSTRACT FROM AUTHOR]

Semi-Supervised Graph Learning (SSGL) aims to predict massive unknown labels based on a subset of known labels within a graph. Recently, graph neural network, one of the most popular SSGL approaches, has garnered considerable research interest and achieved remarkable progress. However, many of these methods perform unsatisfactorily with limited labeled data. Graph contrastive learning (GCL), which utilizes unlabeled data to generate supervision, partially addresses this issue but does not fully exploit label information. To address this challenge, we propose SSGL algorithm, Semi-supervised Graph Contrastive Learning with Confidence Propagation Algorithm (SGCL). SGCL comprises two stages of contrastive learning. In the first stage, we employ unsupervised contrastive learning to initialize the model with graph augmentation. In the second stage, in order to fully leverage known labels and graph structure, we incorporate supervised contrastive learning which utilizes supervision signals obtained from confidence propagation algorithm. By combining supervised contrastive learning and unsupervised contrastive learning, the embedding quality and the classification accuracy can be further enhanced. At last, comprehensive experiments demonstrate that SGCL outperforms the best baseline method by an average of 2.23% across six datasets, highlighting the effectiveness of our approach. [ABSTRACT FROM AUTHOR]

Contrastive learning is a commonly used framework in the field of graph self-supervised learning, where models are trained by bringing positive samples closer together and pushing negative samples apart. Most existing graph contrastive learning models divide all nodes into positive and negative samples, which leads to the selection of some meaningless samples and reduces the model's performance. Additionally, there is a significant disparity in the ratio between positive and negative samples, with an excessive number of negative samples introducing noise. Therefore, we propose a novel dynamic sampling strategy that selects more meaningful samples from the perspectives of structure and features and we incorporate an iteration-based sample selection process into the model training to enhance its performance. Furthermore, we introduce a curriculum learning training method based on the principle of starting from easy to difficult. Sample training for each iteration is treated as a task, enabling the rapid capture of relevant and meaningful sample information. Extensive experiments have been conducted to validate the superior performance of our model across nine real-world datasets. [ABSTRACT FROM AUTHOR]

The long-tail distribution of items is common in recommendation systems. However, due to the limited interaction records of long-tail items, recommending them to users significantly affects the model's performance. Hence, to address this issue, recent research utilized data from popular/head items to supplement data from long-tail items, including transfer learning, meta-learning, and contrastive learning. While these methods have been effective, they still suffer from two challenges: (1) The knowledge transferred to long-tail items is not well-adapted, and (2) the feature representations of long-tail items are not accurate enough. Therefore, this paper combines graph contrastive learning and zero-shot learning to generate virtual feature representations for addressing the data sparsity problem. This novel learning scheme, GACRec, is based on a generative adversarial network (GAN) to generate virtual feature representations for long-tail items. These virtual representations train the model to obtain robust generalization ability. Specifically, graph contrastive learning firstly trains the features of popular and long-tail items. Then, the common interaction records of popular and long-tail items are extracted as shared attributes. Finally, we utilize generative adversarial zero-shot learning to generate virtual representations based on the shared attributes. These virtual representations are then used to replace the feature representations of long-tail items during model training. Through theoretical and experimental analyses, we demonstrate that GACRec improves the model's generalization ability and recommendation accuracy. Extensive experiments on three benchmark datasets demonstrate the effectiveness of GACRec in terms of Precision, Recall, and NDCG. Furthermore, the experimental results highlight that the proposed method outperforms other comparative methods. • A novel zero-shot learning based recommendation framework is proposed. • Nodes and edges are added to the graph structure to enhance sparse data representation. • The augmented representations are used in self-supervised contrastive learning to obtain better feature representations for long-tail items. [ABSTRACT FROM AUTHOR]

Ever-evolving healthcare applications have witnessed a surge in the utilization of electronic health records (EHR) for predicting future patient diagnoses. While Graph Neural Networks have demonstrated that promise in modeling disease-patient relationships, challenges arise from the sparsity and imbalance of patient and diagnostic data. Moreover, the existing models face difficulties in learning the unique disease combination features of patients. To address these challenges, we proposed a novel disease. prediction architecture based on Contrastive Learning (CL) from interstructural and intrasemantic perspectives, rather than traditional CL methods. We generated an initial global static disease graph to directly represent the relationships. among all diseases and a local dynamic disease graph to capture the indirect latent disease relationships among different patients. Multiple CL tasks were designed to learn sparse and imbalanced potentials. Relationships Between Diseases. Interstructure graph CL was first proposed to sample a graph enhancement, based on the distribution of nodes in the global disease graph. To further explore the deep embedding space of the disease, an intra-view graph CL was introduced by injecting noise at the semantic level for robust graph comparison. Experimental validation on two real EHR datasets demonstrates the superior performance of the approach by comparing it with state-of-the-art models. • Multi-view contrastive learning to address data sparsity in disease prediction. • Enhancing disease embeddings by structural and semantic contrastive learning views. • Global and local graphs synergistically extract complex patient-disease relationships. • Mining complex disease hierarchies through direct and indirect neighbor relationships in graphs. [ABSTRACT FROM AUTHOR]

Detecting events from social media data streams is gradually attracting researchers. The innate challenge for detecting events is to extract discriminative information from social media data thereby assigning the data into different events. Due to the excessive diversity and high updating frequency of social data, using supervised approaches to detect events from social messages is hardly achieved. To this end, recent works explore learning discriminative information from social messages by leveraging graph contrastive learning (GCL) and embedding clustering in an unsupervised manner. However, two intrinsic issues exist in benchmark methods: conventional GCL can only roughly explore partial attributes, thereby insufficiently learning the discriminative information of social messages; for benchmark methods, the learned embeddings are clustered in the latent space by taking advantage of certain specific prior knowledge, which conflicts with the principle of unsupervised learning paradigm. In this paper, we propose a novel unsupervised social media event detection method via hybrid graph contrastive learning and reinforced incremental clustering (HCRC), which uses hybrid graph contrastive learning to comprehensively learn semantic and structural discriminative information from social messages and reinforced incremental clustering to perform efficient clustering in a solidly unsupervised manner. We conduct comprehensive experiments to evaluate HCRC on the Twitter and Maven datasets. The experimental results demonstrate that our approach yields consistent significant performance boosts. In traditional incremental setting, semi-supervised incremental setting and solidly unsupervised setting, the model performance has achieved maximum improvements of 53%, 45%, and 37%, respectively. [Display omitted] • A novel unsupervised social event detection model, namely HCRC. • Effective approach for building hybrid social message graphs. • Solidly unsupervised incremental clustering module using reinforcement learning. • Comprehensive validation of HCRC through three distinct experimental settings. [ABSTRACT FROM AUTHOR]

Anomaly detection using multivariate time series plays a crucial role in system security. Conventional deep learning detection techniques mainly depend on temporal dependency and employ reconstruction or prediction-based methods. However, as feature variables grow more intricate, there is a risk of neglecting essential spatio-temporal structural information, potentially leading to insufficient model training in unsupervised settings. Hence, we propose an end-to-end anomaly detection model with multiple pre-training tasks designed for the spatio-temporal dimension to enhance our constraints. Specifically, in the temporal dimension, we employ an autoregressive task to train timestamp associations using data's concealed autocorrelation and periodicity. In the spatio dimension, we acquire knowledge of a diverse feature-related heterogeneous graph. Subsequently, we design three different graph contrastive learning tasks to tap into the effective information arising from the inherent heterogeneity and hierarchy in spatio structures. Through joint spatio-temporal modeling, we can effectively capture inter and intra-feature associations from series and graph structural features, enhancing model robustness to cope with the complex chain reactions between features. Finally, we assess our model on three real-world datasets: SWaT, WADI(2017, 2019), our F1 scores demonstrate enhancements of 6.17%, 18.3% and 5.35% over the top-tier baseline performance. Our model is applicable for both temporal and graph, is self-supervised learning for sparse data which is suitable for data sparsity and complex scenarios that need to capture spatio-temporal characteristics at the same time, for example, traffic flow detection and anomaly detection of intelligent systems. Further visualization experiments and case studies will provide a better interpretation of our model. [ABSTRACT FROM AUTHOR]

Graph contrastive learning aims to learn informative and discriminative node representations for downstream tasks by maximizing the mutual information between representations of different augmentation views of the same node. However, according to the multi-view information bottleneck principle, redundant information in learned representations can negatively impact the performance of downstream tasks. To avoid this issue, we propose Adversarial Cluster-Level and Global-Level Graph Contrastive Learning (ACG-GCL) for learning minimal sufficient node representations. ACG-GCL is optimized alternately under an adversarial learning framework with a min–max objective. At the min step, ACG-GCL eliminates redundant information in both the graph structure and feature content of nodes while producing a new graph that provides multi-view information. At the max step, ACG-GCL seeks to preserve shared task-relevant information in the learned representation by maximizing the mutual information of node-cluster level and node-global level representations. We demonstrate the effectiveness of the proposed method on the tasks of node classification and node clustering tasks. Code is available at https://github.com/tangq123/ACG-GCL. [ABSTRACT FROM AUTHOR]

Exploring complicated dynamic spatiotemporal correlations has always been a challenging issue in traffic prediction. Besides, methods that make predictions directly from data with missing values, have received much attention due to the inevitable and pervasive nature of data incompleteness in real scenarios. In this paper, an end-to-end representation learning framework, named spatial–temporal periodical adaptive graph contrastive learning (ST-A-PGCL), is proposed to address such issues. ST-A-PGCL mainly consists of three independent branches to respectively model three long-term periodicities of traffic flow (recent, daily, and weekly periodicities). In each branch, the spatial and temporal correlations are extracted by improved adaptive graph convolution network (ImpAdapGCN) and fused seasonal-trend temporal convolution network (FST-TCN) in an encoder, respectively, to obtain hidden representation. Besides, each branch accepts one periodic segment which will be synthetically augmented with different missing patterns to simulate real scenarios (weak communication signal, detector malfunction, area-wide power failure, etc) and generate different views. These views will be fed into a periodical graph contrastive learning (PGCL) module to learn periodical similarity features based on Siamese network to defeat data incompleteness. Bidirectional gate recurrent unit (Bi-GRU) is selected to decode the hidden representations and generate final prediction results. Specifically, the overall framework is trained in end-to-end dual-task (traffic prediction and contrastive learning) process without requiring identifying the position of missing values. Our framework is evaluated across four real-world datasets and twenty baseline models. Experimental results show that the proposed ST-A-PGCL achieves superior prediction performance, especially in long-term prediction tasks with high missing rates. [ABSTRACT FROM AUTHOR]

With the advent of Web service technologies, the number of services published on the cloud is increasing rapidly. The quality of service (QoS) becomes a crucial criterion for selecting services from a massive pool of candidates. Collaborative filtering (CF) has become a major way for personalized QoS prediction by leveraging historical interactions between users and services. Due to the increasing number of users and services, CF-based QoS prediction often suffers from data sparsity and cold-start difficulties. Inspired by the advantages of graph contrastive learning in cold-start predictions, we propose BGCL, a bi-subgraph network based on graph contrastive learning to solve the above problems. Firstly, we generate different perspectives of user-neighborhood and service-neighborhood sub-graphs based on sparse user–service bipartite graphs. Next, our model learns user and service embeddings using the graph contrastive learning and graph attention aggregation mechanisms on the generated sub-graphs. Finally, user and service embeddings are fed into a multi-layer perception to predict QoS values. Experimental results show that our model outperforms several existing models in terms of prediction accuracy. [ABSTRACT FROM AUTHOR]

Graph contrastive learning (GCL), aiming to generate supervision information by transforming the graph data itself, is increasingly becoming a focus of graph research. It has shown promising performance in graph representation learning by extracting global-level abstract features of graphs. Nonetheless, most GCL methods are performed in a completely unsupervised manner and would get unappealing results in balancing the multi-view information of graphs. To alleviate this, we propose a Semi-supervised Multi-view Graph Contrastive Learning (SMGCL) framework for graph classification. The framework can capture the comparative relations between label-independent and label-dependent node (or graph) pairs across different views. In particular, we devise a graph neural network (GNN)-based label augmentation module to exploit the label information and guarantee the discrimination of the learned representations. In addition, a shared decoder module is complemented to extract the underlying determinative relationship between learned representations and graph topology. Experimental results on graph classification tasks demonstrate the superiority of the proposed framework. [ABSTRACT FROM AUTHOR]