Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
September 30, 2025Open Access

URL2Graph++: Unified Semantic-Structural-Character Learning for Malicious URL Detection

View Full Paper
Ask AI
Bookmark
Share

Authors

YTYe TianNorth University of ChinaYJYifan JiaBeijing University of Posts and TelecommunicationsYWYanbin WangShenzhen University

Discussion

Loading...

Member takes

Implication

Proposed method combines graph learning and semantic embedding for improved malicious URL detection performance.

Key Points

  • Our approach achieves superior detection accuracy, capturing both structural and semantic features.
  • With dual-granularity URL graphs, the model learns internal dependencies to enhance detection capabilities.
  • Using BERT enhances semantic understanding, boosting performance against sophisticated obfuscation techniques.
  • Results demonstrate a marked improvement over state-of-the-art methods, including those based on large language models.

Cite This Study

Tian et al. (2025) studied this question.

synapsesocial.com/papers/68dc1e358a7d58c25ebb1823https://doi.org/10.48550/arxiv.2509.10287
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1WebGuard++:Interpretable Malicious URL Detection via Bidirectional Fusion of HTML Subgraphs and Multi-Scale Convolutional BERT2025
  2. 2Cyber-Attack Detection of Malicious URLs Using Deep-Learning Techniques2026
  3. 3Multimodal fusion for malicious URL detection using visual, structural, and semantic representations2026
  4. 4PLG-URLNet: Layer-Wise Fusion of Local Convolution and Global Attention for Detecting Malicious URLs2026
  5. 5Lightweight malicious URL detection using deep learning and large language models2025