GRID: Graph Representation of Intelligence Data for Security Text Knowledge Graph Construction
The article discusses a new framework called GRID for constructing security text knowledge graphs from cyber threat intelligence. It addresses challenges in using large language models for this task and presents a method for improving the precision and recall of the resulting graphs. The framework demonstrates significant performance improvements over existing methods in terms of stability and efficiency.
- ▪GRID is an end-to-end framework designed for security text knowledge graph construction.
- ▪The framework utilizes a task bank approach to provide more stable rewards compared to traditional methods.
- ▪The Task-bank Reward model achieved 84.62% source-averaged precision and 64.91% source-averaged recall on a dataset of 249 CTI articles.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.16714 (cs) [Submitted on 15 May 2026] Title:GRID: Graph Representation of Intelligence Data for Security Text Knowledge Graph Construction Authors:Liangyi Huang, Zichen Liu, Fei Shao, Shang Ma, Mengshi Zhang, Zihao Chen, Yanfang Ye, Xusheng Xiao View a PDF of the paper titled GRID: Graph Representation of Intelligence Data for Security Text Knowledge Graph Construction, by Liangyi Huang and 7 other authors View PDF HTML (experimental) Abstract:Security knowledge graphs can provide computable external memory for security agents, but constructing them from long-form cyber threat intelligence (CTI) remains difficult: LLMs often lack grounded security-domain knowledge, and end-to-end document-to-graph training is hard to supervise with…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
