Arrow Research search

Author name cluster

Fangyu Ding

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

2 papers
1 author row

Possible papers

2

EAAI Journal 2026 Journal Article

Predicting inter-state cyberattacks with graph-text fusion using graph neural networks and large language models

  • Jiping Dong
  • Mengmeng Hao
  • Fangyu Ding
  • Shuai Chen
  • Jiajie Wu
  • Jun Zhuo

Accurate forecasting of inter-state cyberattacks is crucial for the timely prevention of security risks. However, the dynamic evolution of international relations and the heterogeneity of multi-source data make this task significant challenging in both data integration and model design. To address these issues, we propose GeoDyG-LLM (Geopolitical Dynamic Graph-Large Language Model), a unified multimodal framework for geopolitically grounded cyberattack prediction. The framework models inter-state interactions by integrating dynamic graph neural networks (GNNs) with large language models (LLMs), jointly capturing the temporal, structural, and semantic dependencies from historical cyberattack records and news events. Methodologically, the framework (i) employs a dynamic multi-view GNN with a learnable projector, whose node embeddings are layer-wise injected into the Transformer layers of the LLM; (ii) leverages LLM-based semantic refinement to construct a high-quality multimodal dataset from raw data sources; and (iii) adopts a geopolitically-aware negative sampling strategy to generate informative and balanced training pairs. Experimental results show that GeoDyG-LLM (8B) substantially outperforms baselines, achieving an F1 score of 0. 888 on cyberattack prediction. Moreover, the fine-tuned model generates coherent natural-language explanations aligned with historical and contextual evidence, enhancing interpretability and supporting practical geopolitical cyber risk analysis.

Details DOI

AAAI Conference 2023 Conference Paper

C-NTPP: Learning Cluster-Aware Neural Temporal Point Process

  • Fangyu Ding
  • Junchi Yan
  • Haiyang Wang

Event sequences in continuous time space are ubiquitous across applications and have been intensively studied with both classic temporal point process (TPP) and its recent deep network variants. This work is motivated by an observation that many of event data exhibit inherent clustering patterns in terms of the sparse correlation among events, while such characteristics are seldom explicitly considered in existing neural TPP models whereby the history encoders are often embodied by RNNs or Transformers. In this work, we propose a c-NTPP (Cluster-Aware Neural Temporal Point Process) model, which leverages a sequential variational autoencoder framework to infer the latent cluster each event belongs to in the sequence. Specially, a novel event-clustered attention mechanism is devised to learn each cluster and then aggregate them together to obtain the final representation for each event. Extensive experiments show that c-NTPP achieves superior performance on both real-world and synthetic datasets, and it can also uncover the underlying clustering correlations.

PDF Details DOI
v2026.09.13