Multimodal named entity recognition (MNER) aims to identify predefined entity types from text by incorporating auxiliary information such as images. Compared with traditional named entity recognition methods, MNER introduces visual context to recognize entities in ambiguous text. To more effectively utilize the visual modality for text understanding, most current approaches rely on static cross-modal feature alignment and fusion. However, when text-image correlations are weak, such strategies fail to align the semantic information between modalities. In addition, existing methods tend to model local and global semantic representations independently, lacking a unified framework that can jointly capture both local and global semantic information. To address these issues, we propose a unified semantic alignment and cross-graph fusion model framework (SAGF) for MNER. First, we introduce a multimodal adaptive semantic alignment mechanism that uses a trainable bilinear similarity function to adaptively align cross-modal semantic information under weak text-image correlation. Second, we propose a cross-modal graph fusion strategy. It combines local semantic representations based on attention mechanisms, with global structure features encoded in a unified graph structure. This joint local with global modeling enables the framework to capture semantic relations and contextual structures among entities. Experiments on two well-known social media datasets demonstrate the effectiveness of the SAGF model, achieving strong performance in MNER.

Network embedding (NE) focuses on discovering low-dimensional embeddings of nodes while retaining their intrinsic features and structure of nodes. It is essential for many practical applications, containing text mining, community detection, and node classification. However, the great majority of existing systems are incapable of combining structural and attribute information. To tackle the above-mentioned problem, considering the information diffusion process, we present a novel model for attribute NE (ANE), namely influential node diffusion-based matrix factorization (INDMF), which contains topology level and attribute level. In detail, we first propose a novel method to extract high-order information via influential node diffusion sequences. Then, we regard the optimization of our proposed structure-based and attribute-based loss functions as a matrix factorization problem. Furthermore, this model can be used to generate final node embedding by aggregating the topology level and attribute level hierarchically. Experiments are conducted on four real-world datasets, which indicates that INDMF beats all competing algorithms in node categorization, community detection, and graph visualization.

Influence maximization (IM) aims to find a subset of$k$nodes that can maximize the final active node set under an information diffusion model. With the development and popularity of social networks, the IM problem plays an essential role in various applications, such as public opinion analysis, viral marketing, and rumor early warning. However, most of the existing IM solutions have not accessed the risk of negative information from nodes in the future. In reality, a company may face economic loss when negative information breaks out of its spokesman. Therefore, a crisis assessment oriented and topic-based IM problem (TIM-CA) is proposed, which is utilized to model the IM problem by considering the crisis assessment (CA) and topics of users. To solve this problem, we propose a maximum influence arborescence model-based algorithm for TIM-CA, namely, MIA-TIM-CA. The proposed algorithm consists of crisis degree calculation, topic relevance calculation, and influence spread evaluation. More importantly, as for crisis degree calculation, it considers self-, topology-, and topic-based crisis degrees for each node. At the influence spread evaluation stage, MIA-TIM-CA proposes two functions to evaluate the node’s importance and node influence spread. Extensive experiments on two real-world social networks demonstrate that our MIA-TIM-CA outperforms all comparison algorithms on influence spread, crisis score, and running time.

Network embedding (NE) focuses on mapping each node within a network into a condensed vector representation of lower dimensional, while simultaneously retaining the original network structure and attribute features as faithfully as possible. It applies to a wide range of tasks of network mining, including link prediction, node classification, user recommendation, etc. Most existing network embedding models primarily aim at static networks, which require node knowledge in the full-time span. However, various real-world network is dynamic, and the nodes in the network change frequently. Although dynamic network embedding has garnered attention from a limited number of researchers, their approach has been restricted to learning node representations in isolation, without considering the integration of higher-order topological structure and attribute features of the nodes. To better address the above problems, a Temporal Evolution Network Embedding Model based on Structural and Attributes Retention (TNSAR) is proposed. In summary, our contributions include Introducing the graph decomposition technique for capturing global higher-order topological features, including higher-order structure, a fusion of attribute and topological information in the Graph Convolutional Network (GCN) network for node representation, and the temporal preservation component. The experimental results validate the effectiveness of TNSAR in network embedding tasks, showcasing its superiority over baseline methods.

Knowledge graph (KG) completion aims to infer new facts from incomplete knowledge graphs. Most existing solutions focus on learning from time-aware fact triples and ignore the spatial information. In reality, knowledge graphs can evolve with time as well as the changing locations, such as the flight domain. Therefore, integrating spatiotemporal information into knowledge graph representation is important for the knowledge graph completion. To address this problem, this paper proposes two Spatio Temporal-aware knowledge graph completion models based on the Sequence Encoder, namely STSE and S-TSE, which incorporate the spatial and temporal information into relations. Specifically, the model consists of two steps: spatiotemporal-aware relation encoding and final scoring function evaluation. The first stage composes the spatiotemporal information into different tokens. Then two methods are proposed to obtain the embedding of spatiotemporal-aware relation by utilizing the Recursive Neural Network. The second stage proposes different scoring functions for two models. Empirically evaluation of the proposed models is conducted on spatiotemporal-aware KG completion task on two public datasets. Experimental results demonstrate the effectiveness of the proposal for spatiotemporal knowledge graph completion.

The purpose of the influence maximization is to find the top k influential seeds which can maximize the influence spread. Recently, some researchers address this problem through community structure. However, most of these community-based studies only consider the static social network, which ignore s that social networks change frequently. In order to deal with the above problem, we present a community-based algorithm on the dynamic social network, which is divided into three phases: (i) community detection, (ii) candidate seed set on the dynamic network, and (iii) final seed set. In the first phase, we use the Louvain algorithm to obtain the community structure. In each community, we analyze the node location to judge the importance of the node. In the second phase, considering the dynamic social network, when a node is added to the network or removed from the network, we update the structure of the social network. Then, the candidate nodes are those nodes with a large influence in each community. And in the third phase, we select k influential seeds from the candidate seeds by CELF algorithm. Extensive experimental results show that our algorithm obtains a better influence spread than many baseline algorithms as well as an acceptable running time while considering the dynamic social networks.