Graph Data Research Articles

MA-GCL4SR: Improving Graph Contrastive Learning-Based Sequential Recommendation with Model Augmentation

Sequential recommendation (SR) has leveraged the advantages of graph contrastive learning (GCL) to enhance the representation of SR, which mitigates to some extent the constraint of scarce labeled data for supervision in SR. Existing work applies general graph data augmentation strategies to generate positive sample pairs, then further representation learning is conducted through a shared graph neural network. In this study, we identify limitations in applying traditional GCL to sequential recommendation: after the data augmentation, the shared graph neural network architecture used for feature learning fails to supply sufficiently diverse contrastive views, which are necessary to effectively identify and focus on the key information that is truly relevant for sequential recommendation. To ease this limitation, we propose a novel framework named Model Augmented Graph Contrastive Learning for Sequential Recommendation (MA-GCL4SR), which emphasizes modifying the internal architectures of the graph neural network through the use of model augmentation strategies, rather than focusing on making improvements during the data augmentation phase before encoding. Thereby, we construct a non-shared view encoder for SR, enriching the samples of user’s interaction sequences and strengthen the stability of the augmented sequence. Extensive experiments on four real-world datasets confirm the effectiveness of the proposed MA-GCL4SR paradigm, showcasing its consistent ability to elevate model performance across various real-world scenarios.

Graph-to-Text Generation with Bidirectional Dual Cross-Attention and Concatenation

Graph-to-text generation (G2T) involves converting structured graph data into natural language text, a task made challenging by the need for encoders to capture the entities and their relationships within the graph effectively. While transformer-based encoders have advanced natural language processing, their reliance on linearized data often obscures the complex interrelationships in graph structures, leading to structural loss. Conversely, graph attention networks excel at capturing graph structures but lack the pre-training advantages of transformers. To leverage the strengths of both modalities and bridge this gap, we propose a novel bidirectional dual cross-attention and concatenation (BDCC) mechanism that integrates outputs from a transformer-based encoder and a graph attention encoder. The bidirectional dual cross-attention computes attention scores bidirectionally, allowing graph features to attend to transformer features and vice versa, effectively capturing inter-modal relationships. The concatenation is applied to fuse the attended outputs, enabling robust feature fusion across modalities. We empirically validate BDCC on PathQuestions and WebNLG benchmark datasets, achieving BLEU scores of 67.41% and 66.58% and METEOR scores of 49.63% and 47.44%, respectively. The results outperform the baseline models and demonstrate that BDCC significantly improves G2T tasks by leveraging the synergistic benefits of graph attention and transformer encoders, addressing the limitations of existing approaches and showcasing the potential for future research in this area.

Graph Distillation with Network Symmetry

Abstract Graph Neural Networks (GNNs) have demonstrated excellent performance in graph representation learning. However, as the volume of graph data grows, issues related to cost and efficiency become increasingly prominent. Graph distillation methods address this challenge by extracting a smaller, reduced graph, ensuring that GNNs trained on both the original and reduced graphs show similar performance. Existing methods, however, primarily optimize the feature matrix of the reduced graph and rely on correlation information from GNNs, while neglecting the original graph's structure and redundant nodes. This often results in a loss of critical information within the reduced graph. To overcome this limitation, we propose a graph distillation method guided by network symmetry. Specifically, we identify symmetric nodes with equivalent neighborhood structures and merge them into "super nodes," thereby simplifying the network structure, reducing redundant parameter optimization, and enhancing training efficiency. At the same time, instead of relying on the original node features, we employ gradient descent to match optimal features that align with the original features, thus improving downstream task performance. Theoretically, our method guarantees that the reduced graph retains the key information present in the original graph. Extensive experiments demonstrate that our approach achieves significant improvements in graph distillation, exhibiting strong generalization capability and outperforming existing graph reduction methods.

An end-to-end explainability framework for spatio-temporal predictive modeling

The rising adoption of AI models in real-world applications characterized by sensor data creates an urgent need for inference explanation mechanisms to support domain experts in making informed decisions. Explainable AI (XAI) opens up a new opportunity to extend black-box deep learning models with such inference explanation capabilities. However, existing XAI approaches for tabular, image, and graph data are ineffective in contexts with spatio-temporal data. In this paper, we fill this gap by proposing a XAI method specifically tailored for spatio-temporal data in sensor networks, where observations are collected at regular time intervals and at different locations. Our model-agnostic masking meta-optimization method for deep learning models uncovers global salient factors influencing model predictions, and generates explanations taking into account multiple analytical views, such as features, timesteps, and node locations. Our qualitative and quantitative experiments with real-world forecasting datasets show that our approach effectively extracts explanations of model predictions, and is competitive with state-of-the-art approaches.

Graph Foundation Models: Concepts, Opportunities and Challenges.

Foundation models have emerged as critical components in a variety of artificial intelligence applications, and showcase significant success in natural language processing and several other domains. Meanwhile, the field of graph machine learning is witnessing a paradigm transition from shallow methods to more sophisticated deep learning approaches. The capabilities of foundation models in generalization and adaptation motivate graph machine learning researchers to discuss the potential of developing a new graph learning paradigm. This paradigm envisions models that are pre-trained on extensive graph data and can be adapted for various graph tasks. Despite this burgeoning interest, there is a noticeable lack of clear definitions and systematic analyses pertaining to this new domain. To this end, this article introduces the concept of Graph Foundation Models (GFMs), and offers an exhaustive explanation of their key characteristics and underlying technologies. We proceed to classify the existing work related to GFMs into three distinct categories, based on their dependence on graph neural networks and large language models. In addition to providing a thorough review of the current state of GFMs, this article also outlooks potential avenues for future research in this rapidly evolving domain.

A Multivariate Spatiotemporal Feature Fusion Network for Wind Turbine Gearbox Condition Monitoring

SCADA data, due to their easy accessibility and low cost, have been widely applied in wind turbine gearbox condition monitoring. However, the high-dimensional and nonlinear nature of the collected data, along with the insufficient spatiotemporal feature capabilities of existing methods and the lack of consideration of the physical mechanisms of wind turbine operation, limit the accuracy of monitoring models. In this paper, a multivariate spatiotemporal feature fusion network is proposed for wind turbine gearbox condition monitoring. First, by analyzing the operational mechanism of wind turbines and the correlation between sensor data, the time series data are transformed into graph data. Then, graph convolutional networks and temporal convolutional networks are used to extract spatial and temporal features, respectively. Next, long short-term memory networks are employed to fuse the extracted temporal and spatial features, further capturing long-term spatiotemporal dependencies. Finally, the proposed method is validated using real data from two wind turbines. Experimental results show that the proposed method reduces the RMSE by 29.67% and 17.61% compared to the best-performing models. Moreover, the proposed method provides early warning signals 188.6 h and 133.67 h in advance, achieving stable and efficient early anomaly detection for wind turbines.

A comparative evaluation of social network analysis tools: performance and community engagement perspectives

Graphs are increasingly used in research, industry, and government. This has led to a wide range of analytical and graph-processing tools. There are various tools and platforms for graph processing. Over time, diverse systems have emerged, employing various statistics and criteria to evaluate their effectiveness and usability in processing graph data. As a result, comparing the various systems’ performances becomes challenging. This study benchmarks popular network analysis tools—NetworkX, RustworkX, Igraph, EasyGraph, and Graph-tool—by evaluating their performance and extracted community engagement metrics, such as the number of downloads, stars, and forks which reflect the tools’ adoption, popularity, and community support. We benchmark the library’s performance on four open-source datasets, one custom dataset, and twelve network analysis methods. The findings reveal that while NetworkX is highly popular, it exhibits slower performance in most benchmarks compared to Graph-tool and Igraph, which are faster and more efficient despite their lower popularity. The continued popularity of NetworkX may be attributed to factors like well-documented methods and a user-friendly API, though this warrants further investigation. This research provides valuable insights for practitioners, researchers, and developers, helping them make informed decisions when selecting network analysis tools. The study emphasizes the trade-off between user-friendliness and performance, suggesting that the optimal tool choice depends on project-specific requirements.

Federated Graph Neural Networks: Overview, Techniques, and Challenges.

Graph neural networks (GNNs) have attracted extensive research attention in recent years due to their capability to progress with graph data and have been widely used in practical applications. As societies become increasingly concerned with the need for data privacy protection, GNNs face the need to adapt to this new normal. Besides, as clients in federated learning (FL) may have relationships, more powerful tools are required to utilize such implicit information to boost performance. This has led to the rapid development of the emerging research field of federated GNNs (FedGNNs). This promising interdisciplinary field is highly challenging for interested researchers to grasp. The lack of an insightful survey on this topic further exacerbates the entry difficulty. In this article, we bridge this gap by offering a comprehensive survey of this emerging field. We propose a 2-D taxonomy of the FedGNN literature: 1) the main taxonomy provides a clear perspective on the integration of GNNs and FL by analyzing how GNNs enhance FL training as well as how FL assists GNN training and 2) the auxiliary taxonomy provides a view on how FedGNNs deal with heterogeneity across FL clients. Through discussions of key ideas, challenges, and limitations of existing works, we envision future research directions that can help build more robust, explainable, efficient, fair, inductive, and comprehensive FedGNNs.

Self-Supervised Anomaly Detection With Neural Transformations.

Data augmentation plays a critical role in self-supervised learning, including anomaly detection. While hand-crafted transformations such as image rotations can achieve impressive performance on image data, effective transformations of non-image data are lacking. In this work, we study learning such transformations for end-to-end anomaly detection on arbitrary data. We find that a contrastive loss - which encourages learning diverse data transformations while preserving the relevant semantic content of the data - is more suitable than previously proposed losses for transformation learning, a fact that we prove theoretically and empirically. We demonstrate that anomaly detection using neural transformation learning can achieve state-of-the-art results for time series data, tabular data, text data and graph data. Furthermore, our approach can make image anomaly detection more interpretable by learning transformations at different levels of abstraction.

OpenFMR: An open-source broadband ferromagnetic resonance spectrometer.

We describe a broadband ferromagnetic resonance spectrometer for scientific and educational applications with a frequency range of up to 30GHz. It is built with components available off-the-shelf, utilizes 3D printed parts for sample holders and support structures, and requires little assembly. A PCB design for the grounded coplanar waveguide (GCPW) is presented and analyzed. We further include a software suite for command-line or script driven data acquisition, a graphical user interface, and a graphical data analysis program. The capabilities of the system design are demonstrated with measurements on ferromagnetic thin films with a thickness of 1nm. All designs and scripts are published under the GNU GPL v3.0 license.

Efektivitas Aplikasi Marbel Berbasis Android untuk Meningkatkan Kemampuan Membaca Permulaan pada Anak Disleksia: Single Subject Research di SDN 28 Batipuh, Kabupaten Tanah Datar

Difficulties in beginning reading skills in second-grade students, especially those who show characteristics of dyslexia, are still a challenge in basic education. This study aims to evaluate the effectiveness of using Marbel application in improving beginning reading skills in dyslexic children. The approach used was a quantitative method with Single Subject Research (SSR) A-B-A experimental design. Data were obtained through a reading test adapted from an instrument developed by Sonia Triandani Yolanda and Damri (2022), with a reliability coefficient of 0.979. The research subjects were second-grade students of SDN 28 Batipuh. The results of the visual analysis of graph data show that in the initial baseline phase (A1), the subject has a stable reading ability level of 13% for five observations. In the intervention phase (B) which lasted for six observations, there was an increase in reading ability, with a stable achievement at 87%. Furthermore, in the final baseline phase (A2), after four observations, reading ability further increased to reach 92%. The findings show that Marbel app is an effective learning media in improving beginning reading skills for children with dyslexia at the primary school level.

Deep graph clustering via aligning representation learning

Deep graph clustering via aligning representation learning

DropNaE: Alleviating irregularity for large-scale graph representation learning.

DropNaE: Alleviating irregularity for large-scale graph representation learning.

Multivariate analysis of near additions for presbyopic patients at a rural optometry clinic

Background: Presbyopia is often overlooked in refractive error distribution analysis. This article employs multivariate analysis to address this gap, enhancing understanding of means, outliers and variations through graphical data presentation.Aim: To analyse the distribution of near-corrective optical additions for presbyopic patients over 2 years at a rural optometry clinic.Setting: The study was conducted at Sekororo District Hospital, South Africa.Methods: Non-cycloplegic near-refractive error data for presbyopic patients who visited the clinic at the district hospital from January 2018 to December 2019 were extracted from the hospital’s records. The records were randomly divided into two groups for 2018 and 2019. Meridional plots and stereo-pair scatter plots were used to analyse the refractive states for the right (OD) and left (OS) eyes.Results: In the 2018 sample, the clinical means for OD and OS were +1.33 ‒0.32 × 90 and +2.01 ‒0.37 × 77, respectively. Similarly, for the 2019 sample, the clinical means for OD and OS were +2.01 ‒0.32 × 82 and +1.82 ‒0.18 × 95, respectively. The data were not normally distributed, and outliers were present. Sample variances were spherical rather than astigmatic.Conclusion: Deviation from the normality showed that the data for OD and OS were mainly mildly positively skewed. Much of the variation in the refractive state was spherical (the stigmatic) irrespective of the laterality.Contribution: The article makes a valuable contribution to the current understanding of multivariate analysis, in academic training including optometry fraternity, as it pertains to the refractive state of the eyes in a rural optometry clinic.

Automated Retrosynthesis Planning of Macromolecules Using Large Language Models and Knowledge Graphs.

Identifying reliable synthesis pathways in materials chemistry is a complex task, particularly in polymer science, due to the intricate and often nonunique nomenclature of macromolecules. To address this challenge, an agent system that integrates large language models (LLMs) and knowledge graphs is proposed. By leveraging LLMs' powerful capabilities for extracting and recognizing chemical substance names, and storing the extracted data in a structured knowledge graph, the system fully automates the retrieval of relevant literature, extraction of reaction data, database querying, construction of retrosynthetic pathway trees, further expansion through the retrieval of additional literature and recommendation of optimal reaction pathways. By considering the complex interdependencies among chemical reactants, a novel Multi-branched Reaction Pathway Search Algorithm (MBRPS) is proposed to help identify all valid multi-branched reaction pathways, which arise when a single product decomposes into multiple reaction intermediates. In contrast, previous studies are limited to cases where a product decomposes into at most one reaction intermediate. This work represents the first attempt to develop a fully automated retrosynthesis planning agent tailored specially for macromolecules powered by LLMs. Applied to polyimide synthesis, the new approach constructs a retrosynthetic pathway tree with hundreds of pathways and recommends optimized routes, including both known and novel pathways.

Improved sand cat swarm optimization algorithm assisted GraphSAGE-GRU for remaining useful life of engine

With the development of deep learning, the potential for its use in remaining useful life (RUL) has substantially increased in recent years due to the powerful data processing capabilities. However, the relationships and interdependencies of operation parameters in non-Euclidean space are ignored utilizing the current deep learning-based methods during the degradation process for engine. To address this challenge, an improved sand cat swarm optimization-assisted Graph SAmple and aggregate and gate recurrent unit (ISCSO-GraphSage-GRU) is proposed to achieve RUL prediction in this work. Firstly, the maximum information coefficient (MIC) is utilized for describing the interdependent relations of measured parameters. Building on this foundation, the constructed graph data is used as input to GraphSage-GRU so as to overcoming the shortcomings of existing deep learning methods. Additionally, this work proposed an improved sand cat swarm optimization (ISCSO) to improve the predicted performance of GraphSage-GRU, including tent mapping in population initialization and a novel adaptive approach enhance the exploration and exploitation of sand cat swarm optimization. The CMAPSS dataset is used to validate the effectiveness and advancedness of ISCSO-GraphSage-GRU, and the experimental results show that the R2 of the ISCSO-GraphSage-GRU is greater than 0.99, RMSE is less than 6.

Enhancing Small Language Models for Graph Tasks Through Graph Encoder Integration

Small language models (SLMs) are increasingly utilized for on-device applications due to their ability to ensure user privacy, reduce inference latency, and operate independently of cloud infrastructure. However, their performance is often limited when processing complex data structures such as graphs, which are ubiquitous in real-world datasets like social networks and system interactions. Graphs inherently encode intricate structural dependencies, requiring models to effectively capture both local and global relationships. Traditional language models, designed primarily for text data, struggle to address these requirements, leading to suboptimal performance in graph-related tasks. To overcome this limitation, we propose a novel graph encoder-based prompt tuning framework which integrates a graph convolutional network (GCN) with a graph transformer. By leveraging the complementary strengths of the GCN for local structural modeling and the graph transformer for capturing global relationships, our method enables SLMs to effectively process graph data. This integration significantly enhances the ability of SLMs to handle graph-centric tasks while maintaining the efficiency required for resource-constrained devices. The experimental results show that our approach not only improves the performance of SLMs on various graph benchmarks but also achieves results which closely approach the performance of a large language model (LLM). This work highlights the potential of extending SLMs for graph-based applications and advancing the capabilities of on-device artificial intelligence.

Denoising structure against adversarial attacks on graph representation learning

Despite their excellent performance in graph representation learning, graph convolutional networks have been proved to be vulnerable to adversarial perturbations on the connectivity between nodes in an unnoticed manner. In this work, by looking into the impacts of adversarial attacks on graph data, we empirically find that the dominant edge-addition attacks generally increase the heterophily between connected nodes, which will fool the transductive inference models on node classification task. To defend against such attacks, we develop a two-stage denoising method (TSD) that aims at removing possible malicious edges so as to mitigate the heterophily issue introduced by attacks. In particular, after a rough removal of the links that have quite low feature similarity, our method further spots the potentially heterophilious links by predicting node labels with a multi-view labeling consensus. This design is based on assumption that if the label predictions for the same node from two different views of a graph data are consistent, then we have a high chance to acquire the reliable labeling. The experiments demonstrate that by denoising a graph this way, the robustness of graph convolutional networks on node classification task is remarkably improved, compared to several strong competitive robust graph neural network models.

Class-Imbalanced Learning on Graphs: A Survey

Rapid advancement in machine learning is increasing the demand for effective graph data analysis. However, real-world graph data often exhibits class imbalance, leading to poor performance of standard machine learning models on underrepresented classes. To address this, C lass- I mbalanced L earning on G raphs (CILG) has emerged as a promising solution that combines graph representation learning and class-imbalanced learning. This survey provides a comprehensive understanding of CILG’s current state-of-the-art, establishing the first systematic taxonomy of existing work and its connections to traditional imbalanced learning. We critically analyze recent advances and discuss key open problems. A continuously updated reading list of relevant papers and code implementations is available at https://github.com/yihongma/CILG-Papers.

Transient Stability Analysis and Emergency Generator Tripping Control Based on Spatio-Temporal Graph Deep Learning

This paper addresses the challenge of achieving fast and accurate transient stability analysis and emergency control in power systems, which are crucial for reliable grid operation under disturbances. To this end, we propose a spatio-temporal graph deep learning approach leveraging Diffusion Convolutional Gated Recurrent Units (DCGRUs) for transient stability assessment and coherent generator group prediction. Unlike traditional methods, our approach explicitly represents transient responses as spatio-temporal graph data, capturing both topological and dynamic dependencies. The DCGRU model effectively extracts these features, and the predicted coherent generator groups are incorporated into the single-machine infinite-bus equivalence method to design an emergency generator tripping scheme. Simulation analysis results on both benchmark and real-world power grids validate the proposed method’s feasibility and effectiveness in enhancing transient stability analysis and emergency control.