Network Graph Research Articles

Academic Research Trend Analysis in the Past Twenty-five Years of Electric Vehicle Industry by Network Visualisation

The study investigates the visualised network regarding the research focuses and trends of the electric vehicle industry by VOSviewer and Gephi software. A co-occurrence network was constructed based on cleaned data, with a threshold of five and 1000 selected keywords. There are three types of visualisations presented in coloured clusters, overlay and density. The VOSviewer illustrates a network with 991 nodes, 242,749 edges, and a total link strength of 1,884,874. Although three visualisations represent the same network, each contains a unique perspective of information, which is the categorisation of research topics, the time arrangement of the studies over the years, and the influential power. The analysis is based on the network graph, structure, keywords position and parameters. The study finds that research areas of the EV industry are significantly dense and overlapping. The sector is inferred to focus more on innovative EV technologies, which will guide researchers and stakeholders in exploring the field more.

Scalable recurrence graph network for stratifying RhoB texture dynamics in rectal cancer biopsies

The scalable recurrence graph network (SRGNet) is introduced in this paper to improve the accuracy of predicting five-year survival outcomes in rectal cancer patients by analyzing RhoB texture dynamics in biopsies. RhoB, a key biomarker assessed via immunohistochemistry, is crucial in predicting responses to radiotherapy (RT), but variability in staining techniques and tumor heterogeneity often complicate these assessments. SRGNet integrates spatial statistics, nonlinear dynamics, graph theory, and graph convolutional networks to address these challenges. In testing, SRGNet outperformed 10 pre-trained convolutional neural networks, achieving 88% accuracy in biopsies from RT patients, with 67% accuracy for predicting survival under five years and 100% accuracy for survival over five years, along with 100% precision, an F1 score of 0.80, and an AUC of 0.73. For non-RT patients, SRGNet attained 91% accuracy, 100% precision for survival over five years, an F1 score of 0.86, and an AUC of 0.82. These results demonstrate SRGNet’s potential to enhance the precision and reliability of survival predictions in rectal cancer patients, overcoming challenges of RhoB expression variability and tumor heterogeneity.

NHSH: Graph Hybrid Learning with Node Homophily and Spectral Heterophily for Node Classification

Graph Neural Network (GNN) is an effective model for processing graph-structured data. Most GNNs are designed to solve homophilic graphs, where all nodes belong to the same category. However, graph data in real-world applications are mostly heterophilic, and homophilic GNNs cannot handle them well. To address this, we propose a novel hybrid-learning framework based on Node Homophily and Spectral Heterophily (NHSH) for node classification in graph networks. NHSH is designed to achieve state-of-the-art or superior performance on both homophilic and heterophilic graphs. It includes three core modules: homophilic node extraction (HNE), heterophilic spectrum extraction (HSE) and node feature fusion (NFF). More specifically, HNE identifies symmetric neighborhoods of nodes with the same category, extracting local features that reflect these symmetrical structures. Then, HSE uses filters to analyze the high and low-frequency information of nodes in the graph and extract the global features of the nodes. Finally, NFF fuses the above two node features to obtain the final node features in graphs. Moreover, an elaborate loss function drives the network to preserve critical symmetries and structural patterns in the graph. Experiments on eight benchmark datasets validate that NHSH performs comparably or better than existing methods across diverse graph types.

Enhancing Autonomous Vehicle Navigation in Complex Environment With Semantic Proto‐Reinforcement Learning

ABSTRACTDespite great progress in autonomous vehicle (AV) navigation, the technical challenges within this space are still considerable when it comes to successful integration of AVs into complex real‐world environments. To tackle these challenges, this paper presents a new semantic proto‐reinforcement learning (SP‐RL) method for dynamic path planning and real‐time obstacle avoidance of an autonomous car that can be adapted to various weather conditions in localization‐deficient environments, while predicting the human intentions of different shapes on road. This approach seeks to improve navigation capability of AV in dynamic and unstructured environments, as well as to address real‐time detection and avoidance response for obstacles more promptly while being able to adapt its decision‐making system based on weather condition by using the semantic graph network (SGN) within segmentation process therefore enhanced version configured with prototype‐based reinforcement learning (PRL). This innovation is new competitive edge compared previous existing approaches. Dynamic SGN is used to segment challenging 3D and free space environments, so that the AV can comprehend highly unintuitive areas like parking lots, construction site conditionals, or off‐road scenarios. At the same time, PRL is used to help real‐time decision‐making so the AV can quickly and precisely respond unexpected obstacles or changing environments. This approach is confirmed to be effective by extensive testing in the CARLA simulation environment, showing substantial improvement of AV navigation capability. This work demonstrates an exciting step towards solving the most fundamental problems faced by autonomous vehicles and could help to ensure that future AV systems are safer, more robust, and adaptable than current ones. It is appliable for urban areas which represent high volumes of pedestrians and vehicles, industrial sites with changing conditions that may be unpredictable at times or the challenging off‐road areas in rural geographies where typical terrains are rough, uneven, rugged yet not well‐structured. The model is robust to diverse weather conditions that make AV operations more reliable and safer. The model was tested on root mean square error (RMSE), computational time, no crash, obstacles avoidance, and success rate obtaining overall 98%.

Traffic Flow Prediction for Highway Based on Multi-Task Spatiotemporal Graph Network

Traffic Flow Prediction for Highway Based on Multi-Task Spatiotemporal Graph Network

RE-SEGNN: recurrent semantic evidence-aware graph neural network for temporal knowledge graph forecasting

RE-SEGNN: recurrent semantic evidence-aware graph neural network for temporal knowledge graph forecasting

Risk Assessment of Typhoon Disaster Chain Based on Knowledge Graph and Bayesian Network

Typhoon disasters not only trigger secondary disasters, such as rainstorms and flooding, but also bring many negative impacts on the normal operation of urban infrastructure and the safety of people’s lives and property. In order to effectively prevent the risks of typhoon disaster chain, this paper proposes a joint entity and relation extraction model based on RoBERTa-Adv-GPLinker. Then, relying on the ontology theory and methodology, we establish a knowledge graph of typhoon disaster chain. The results show that the joint extraction model based on RoBERTa-Adv-GPLinker outperforms other baseline models in all assessment indexes. Moreover, the constructed knowledge graph of typhoon disaster chain includes secondary disasters and derived disaster impacts. This can largely depict the evolution process of typhoon disaster secondary derivations. On this basis, a risk assessment model of typhoon disaster chain based on Bayesian network is established. Taking Fujian Province as an example, the risk associated with the typhoon disaster chain is assessed, verifying the effectiveness of the method. This study provides a scientific basis for enhancing government emergency response capabilities and achieving sustainable regional development.

History of the Mari Region and local Mari groups in the reflection of the latest Russian historiography: experience of social network modeling

The active growth of the use of information technologies has affected the methodology of historiographical research. This article uses the social network modeling technology for the analysis of the Russian historiography of the 2010s, devoted to various aspects of the Mari Territory and local Mari groups’ history. The information capabilities of the Scientific Electronic Library eLIBRARY.RU which contains the RSCI bibliographic database were used to achieve this aim. A selection of scientific papers on the studied subject matter was formed on the stated resource portal. It includes 627 articles from journals and conference proceedings authored by more than 270 experts in historical and related disciplines. From all the articles a pool of highly cited publications was identified, including 72 scientific papers, on the basis of which a network graph was created in the Gephi program, which allowed to visualize the connections between selected papers. The automated graph stacking helped to make 13 large clusters of publications as well as a number of "peripheral" publications. The article describes in detail the methods and technologies used in the research conduction, shows the general description of the identified "topography" of the stated subject area and its development trends, and describes the most popular research topics at the present stage. The study is novel in that it uses citation data to identify and analyze the structure of communication in this subject area. The conducted social network analysis of the scientific literature has shown that the considered subject area of modern Russian historiography demonstrates a pronounced progress in its development, as evidenced by a significant thematic expansion of research, a large and geographically wide of authors’ corpus, the involvement of a number of new historical sources, the use of new research approaches, as well as significant grant support for research.

Air Corridor Planning for Urban Drone Delivery: Complexity Analysis and Comparison via Multi-Commodity Network Flow and Graph Search

Urban drone delivery, a rapidly evolving sector, holds the potential to enhance accessibility, address last-mile delivery issues, and alleviate ground traffic congestion in cities. Effective Unmanned Aircraft System Traffic Management (UTM) is essential to scale drone delivery. A critical aspect of UTM involves planning a city-wide network with spatially-separated air corridors (air routes). Most existing works have focused on routing problems or air traffic management. Compared to these problems, the air corridor planning problem requires much higher spatial and temporal resolutions and presents computational challenges due to the scale, complexity, and density of urban airspace, along with the coupling issues of multi-path planning. Therefore, we conducted this research to understand the complexity and computational resources required to optimally solve the air corridor planning problem. In this paper, we use a minimum-cost Multi-Commodity Network Flow (MCNF) model, a mathematical model, to model the problem and demonstrate the complexity of air corridor planning through the complexity of MCNF. We then apply Gurobi’s and GLPK’s integer programming (IP) solvers to find optimal solutions. Additionally, we present two existing multi-path graph search algorithms, the Sequential Route Network Planning (SRP) algorithm and the Distributed Route Network Planning (DRP) algorithm, to address this corridor planning problem. Numerical experiments conducted at various scales and settings using IP solvers and graph search algorithms indicate that finding an optimal solution requires significant computational resources and yields only a slight improvement in optimality compared to graph search algorithms. Thus, air corridor planning is complex both theoretically and numerically, and graph search algorithms can provide a feasible solution with good enough optimality for corridor planning in real-world scenarios. Moreover, the multi-path graph search algorithms can easily incorporate side constraints that are known to be impossible to solve with polynomial algorithms, making it more practical for real-world applications. Finally, we demonstrate the application of SRP and DRP in real-world 3D urban scenarios.

A multi-gated deep graph network with attention mechanisms for taxi demand prediction

A multi-gated deep graph network with attention mechanisms for taxi demand prediction

A knowledge-refined hybrid graph model for quality prediction of industrial processes

The complexity of industrial processes has spurred the application of soft sensor techniques for predicting key quality variables based on easy-measurable process variables. Currently, data-driven soft sensors based on Artificial Intelligence techniques have become the mainstream. However, these soft sensing models deeply rely on the quality of training data, where the domain knowledge is often ignored. Meanwhile, a significant amount of labeled data is not fully utilized. To address these issues, this paper proposes a supervised framework based on a knowledge-refined hybrid graph network, which contributes to the artificial intelligence application of nonlinear dynamic soft sensors. The problems of applying traditional artificial intelligence models in soft sensor have been addressed by reconstructing the input module of graph neural networks with knowledge-guided approaches. Both spatial and temporal correlations of process data are captured and the hybrid network significantly improves the reliability and interpretability of the soft sensing model. By incorporating labeled data into the model, the representation of quality information is also enhanced. Finally, the proposed framework was applied to an industrial debutanizer column, and the experimental results fully demonstrate the effectiveness and superiority of the method.

Optimising source code vulnerability detection using deep learning and deep graph network

To enhance the effectiveness of vulnerability detection in software developed using C and C++ programming languages, our study introduces a novel correlation calculation method for analyzing and evaluating Code Property Graphs (CPG). The intelligent computation method proposed in this study comprises three key stages. In the first stage, we present a method for extracting features from the CPG source code. To accomplish this, we integrate three distinct data exploration methods: employing Graph Convolutional Neural (GCN) to extract node features from CPG, utilizing Convolutional Neural Network (CNN) to extract edge features from CPG, and finally employing the Doc2vec natural language processing algorithm to extract source code from CPG nodes. The second stage involves proposing a method for synthesizing CPG source code features. Building on the features acquired in the first stage, our paper introduces a synthesis and construction method to generate feature vectors for the source code. The final stage, stage three, executes the detection of source code vulnerabilities. The experimental results demonstrate that our proposed model in this study achieves higher efficiency compared to other studies, with an improvement ranging from 3% to 4%.

Moving Toward Metaproteogenomics: A Computational Perspective on Analyzing Microbial Samples via Proteogenomics.

Microbial sample analysis has received growing attention within the last decade, driven by important findings in microbiome research and promising applications in the biotechnological field. Modern mass spectrometry-based methodology has been established in this context, providing sufficient sensitivity, resolution, dynamic range, and throughput to analyze the so-called metaproteome of complex microbial mixtures from clinical or environmental samples. While proteomic analyses were previously restricted to common model organisms, next-generation sequencing technologies nowadays allow for the rapid and cost-efficient characterization of whole metagenomes of microbial consortia and specific genomes from non-model organisms to which microbes contribute by significant amounts. This proteogenomic approach, meaning the combined application of genomic and proteomic methods, enables researchers to create a protein database that presents a tailored blueprint of the microbial sample under investigation. This contribution provides an overview of the computational challenges and opportunities in proteogenomics and metaproteomics as of January 2018. For practical application, we first showcase an integrative proteogenomic method that circumvents existing reference databases by creating sample-specific transcripts. The underlying algorithm uses a graph network approach that combines RNA-Seq and peptide information. As a second example, we provide a tutorial for a simulation tool that estimates the computational limits of detecting microbial non-model organisms. This method evaluates the potential influence of error-tolerant searches and proteogenomic approaches on databases of interest. Finally, we discuss recommendations for developing future strategies that may help overcome present limitations by combining the strengths of genome- and proteome-based methods and moving toward an integrated metaproteogenomics approach.

Lightweight Dual-Stream SAR–ATR Framework Based on an Attention Mechanism-Guided Heterogeneous Graph Network

Lightweight Dual-Stream SAR–ATR Framework Based on an Attention Mechanism-Guided Heterogeneous Graph Network

Climate and performance-driven architectural floorplan optimization using deep graph networks

PurposeThis study introduces a novel approach to generating and optimizing energy-efficient and climate-responsive architectural floorplans.Design/methodology/approachThe DGraph-cGAN model utilizes advanced deep-learning techniques to produce diverse, realistic layouts that meet specific design constraints and functional requirements.FindingsThe results show significant energy savings (32.1% overall) across different building types and climate conditions, with reductions in energy use intensity, CO2 emissions and annual energy costs. Case studies demonstrate notable improvements in energy savings, CO2 emission reduction, daylight autonomy, thermal comfort and cost savings.Practical implicationsThe DGraph-cGAN model has great potential for advancing architectural design optimization, with opportunities for further refinement and application in various contexts.Originality/valueThis study contributes to developing a novel approach to optimizing architectural floorplans using deep learning techniques. It provides a valuable tool for architects and designers to create energy-efficient, climate-responsive buildings.

Monophonic Cover Pebbling Number \((MCPN)\) of Network Graphs

Monophonic Cover Pebbling Number \((MCPN)\) of Network Graphs

A Hybrid Model for Soybean Yield Prediction Integrating Convolutional Neural Networks, Recurrent Neural Networks, and Graph Convolutional Networks

Soybean yield prediction is one of the most critical activities for increasing agricultural productivity and ensuring food security. Traditional models often underestimate yields because of limitations associated with single data sources and simplistic model architectures. These prevent complex, multifaceted factors influencing crop growth and yield from being captured. In this line, this work fuses multi-source data—satellite imagery, weather data, and soil properties—through the approach of multi-modal fusion using Convolutional Neural Networks and Recurrent Neural Networks. While satellite imagery provides information on spatial data regarding crop health, weather data provides temporal insights, and the soil properties provide important fertility information. Fusing these heterogeneous data sources embeds an overall understanding of yield-determining factors in the model, decreasing the RMSE by 15% and improving R2 by 20% over single-source models. We further push the frontier of feature engineering by using Temporal Convolutional Networks (TCNs) and Graph Convolutional Networks (GCNs) to capture time series trends, geographic and topological information, and pest/disease incidence. TCNs can capture long-range temporal dependencies well, while the GCN model has complex spatial relationships and enhanced the features for making yield predictions. This increases the prediction accuracy by 10% and boosts the F1 score for low-yield area identification by 5%. Additionally, we introduce other improved model architectures: a custom UNet with attention mechanisms, Heterogeneous Graph Neural Networks (HGNNs), and Variational Auto-encoders. The attention mechanism enables more effective spatial feature encoding by focusing on critical image regions, while the HGNN captures interaction patterns that are complex between diverse data types. Finally, VAEs can generate robust feature representation. Such state-of-the-art architectures could then achieve an MAE improvement of 12%, while R2 for yield prediction improves by 25%. In this paper, the state of the art in yield prediction has been advanced due to the employment of multi-source data fusion, sophisticated feature engineering, and advanced neural network architectures. This provides a more accurate and reliable soybean yield forecast. Thus, the fusion of Convolutional Neural Networks with Recurrent Neural Networks and Graph Networks enhances the efficiency of the detection process.

DYNAMIC GENERATIVE RESIDUAL GRAPH CONVOLUTIONAL NEURAL NETWORKS FOR ELECTRICITY THEFT DETECTION

Illegal electricity users pose a significant threat to the economic and security aspects of the power system by illicitly accessing or manipulating electrical resources. With the widespread adoption of Advanced Metering Infrastructure (AMI), researchers have turned to leveraging smart meter data for electricity theft detection. However, existing models rely on methods that model a single electricity load curve and cannot capture the temporal dependencies, periodicity, and underlying features between electricity consumption cycles. This study introduces a novel electricity theft detection method based on dynamic residual graph networks. Innovatively, it proposes a dynamic topological graph construction technique that allows for the real-time updating of adjacency matrices during the training process, thereby effectively capturing the complex relationships in electricity usage patterns. Utilizing the MixHop graph convolutional network, it delves into the temporal sequence dependencies, periodicity, and hidden characteristics within user electricity consumption data. Additionally, to address the issue of model instability caused by scarce theft data, we employ the SMOTE (Synthetic Minority Over-sampling Technique) oversampling technique and enhance overall classification performance by modifying class weights in the loss function. We trained this network architecture on the real SGCC (State Grid Corporation of China) dataset, and the results demonstrate its superiority over other mainstream existing models.

Elevating Social Network Analysis with a Graph Network and Reinforcement Learning Integration for Node Importance

Elevating Social Network Analysis with a Graph Network and Reinforcement Learning Integration for Node Importance

Detecting anomalies in graph networks on digital markets

Detecting anomalies in graph networks on digital markets