Graph representation learning has re-emerged as a fascinating research topic due to the successful application of graph convolutional networks (GCNs) for graphs and inspires various downstream tasks, such as node classification and link prediction. Nevertheless, existing GCN-based methods for graph representation learning mainly focus on static graphs. Although some methods consider the dynamic characteristics of networks, the global structure information, which helps a node to gain worthy features from distant but valuable nodes, has not received enough attention. Moreover, these methods generally update the features of the nodes by averaging the features of neighboring nodes, which may not effectively consider the importance of different neighboring nodes during the aggregation. In this article, we propose a novel representation learning for dynamic graphs based on the GCNs, called DGCN. More specifically, the long short-term memory (LSTM) is utilized to update the weight parameters of GCN for capturing the global structure information across all time steps of dynamic graphs. Besides, a new Dice similarity is proposed to overcome the problem that the influence of directed neighbors is unnoticeable, which is further used to guide the aggregation. We evaluate the performance of the proposed method in the field of node clustering and link prediction, and the experimental results show a generally better performance of our proposed DGCN than baseline methods.

Robustness is one of the most important performance criteria for any rail transit network (RTN), because it helps us enhance the efficiency of RTN. Several studies have addressed the issue of RTN robustness primarily from the perspectives of given rail network structures or static distributions of passenger flow. An open problem that remains in fully understanding RTN robustness is how to take the spatio-temporal characteristics of passenger travel into consideration, since the dynamic passenger flow in an RTN can readily trigger unexpected cascading failures. This paper addresses this problem as follows: (1) we propose a two-layer rail transit network (TL-RTN) model that captures the interactions between a rail network and its corresponding dynamic passenger flow network, and then (2) we conduct the cascading failure analysis of the TL-RTN model based on an extended coupled map lattice (CML). Specifically, our proposed model takes the strategy of passenger flow redistribution and the passenger flow capacity of each station into account to simulate the human mobility behaviors and to estimate the maximum passenger flow appeal in each station, respectively. Based on the smart card data of RTN passengers in Shanghai, our experiments show that the TL-RTN robustness is related to both external perturbations and failure modes. Moreover, during the peak hours on weekdays, due to the large passenger flow, a small perturbation will trigger a 20% cascading failure of a network. Having ranked the cascade size caused by the stations, we find that this phenomenon is determined by both the hub nodes and their neighbors.

Community structure division is a crucial problem in the field of network data analysis. Algorithms based on Markov chains are easy to use and provide promising solutions for community detection. In a Markov chain-based algorithm (i.e., MCL), a flow distribution matrix and a transition matrix are used to describe stochastic flows and transition probabilities, respectively, on a network. The dynamic interaction process between stochastic flows and transition probabilities in MCLs is manifested through an iterative process of updating the abovementioned two matrices. As one of the key mechanisms of MCLs, such a dynamic process for increasing the inhomogeneity directly affects the accuracy and computational cost of MCL-based methods. Inspired by a kind of positive feedback interaction of a dendritic network of tube-like amoeba cell pseudopodia (named the Physarum foraging network), a Physarum -inspired relationship among vertices is proposed to enhance the transition probability in the dynamic process of MCL-based community detection algorithms. Specifically, the proposed hybrid community detection algorithm can adaptively search for a better combination of parameters based on a genetic algorithm. Some experiments are carried out on both static and dynamic networks. The results show that the unique Physarum inspired algorithm achieved better computational efficiency and detection performance than other algorithms.

With the development of modern technology, numerous economic losses are incurred by various spreading phenomena. Thus, it is of great significance to identify the initial sources triggering such phenomena. The investigation of source localization in social networks has gained substantial attention and become a popular topic of study. For practical spreading phenomena on social networks, the infection rates are relatively low. Hence, a high uncertainty of spreading trace might be incurred, which further incurs the reduction of localization accuracy obtained through existed source localization methods, especially the observer-based ones. Aiming to solve the source localization problem with a low infection rate, we propose a novel localization algorithm, i.e., path-based source identification (PBSI). First, a small number of nodes are selected and designated as observers. After the propagation process triggered by sources, we can obtain a snapshot. Later, a label is assigned to represent whether a node is infected or not, and observers are supposed to record the paths through which nodes are successfully infected. Based on source centrality theory, observers make the labels flow in the direction recorded during the label iteration process, which ensures the labels of nodes in the direction of the source increase gradually. Extensive experiments indicate that the proposed PBSI can handle source localization problems for both single and multi-source scenarios with better performance than that of state-of-the-art algorithms under different propagation models.

In the real world, understanding and discovering community structures of networks are significant in exploring network behaviours and functions. In addition to the effect of the closeness of edges on community detection, the node similarity and the structural similarity of networks, which provide auxiliary representations of a network, are also important factors affecting the accuracy of community detection. In this paper, we first represent two similarities by measuring the degree of closeness between nodes and the similarity between two nodes far apart from each other. Then, such similarities are embeded into the low-dimensional vector space by our proposed structural equivalence embedding method based on the non-negative matrix factorization for community detection (SENMF). Extensive experiments demonstrate the effectiveness of our proposed SENMF method compared with several famous network embedding methods and traditional community detection methods.

The growth of social media platforms has made it easier for fake news to spread, which poses a significant threat to authoritative news outlets, politics, and public health. Manual verification of the massive amount of online information has proven to be a daunting task, which has led to the growing interest in automatic fake news detection. Some methods that rely on news text, images, external knowledge, social contexts, or propagation graphs have demonstrated good performance. In contrast to earlier studies that focused solely on the unimodal news textual information, recent works have integrated multimodal features from various granularities, such as words, visual semantic regions, and multimodal entities, to more effectively leverage news content and align with human reading habits. However, a comprehensive review of Multimodal Fake News Detection (MFND) is still lacking, prompting our aim to complement this topic. Specifically, we present a systematic taxonomy from the perspective of cross-modal interactions. We categorize existing methods into the data-based, entity-based, and knowledge-based approaches. Connections between various works are detailed when outlining representative papers. Additionally, we introduce prevalent multimodal learning methods, present accessible MFND datasets and evaluation metrics, and analyze current research results. Finally, the promising future research directions are discussed.