.png){kind=icon}
.png){kind=icon}
Return
Leveraging large language models for crash causation chain inference with in-depth accident investigation data
.png){kind=icon}
PRE
AI
B
王
F
Y
Y
M
DOI：10.1016/j.trc.2026.105820.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• An LLM-based framework is proposed for case-specific crash causation-chain inference. • The Road-Crash-Causation-Chain dataset is a curated dataset constructed for supervised fine-tuning on case-specific DREAM causation-chain inference. • A task-specific evaluation metric was developed for structured causation-chain outputs. • Evaluation on three open-source LLM backbones demonstrates the effectiveness of the proposed framework for crash causation-chain inference tasks.
Keywords:
Road traffic safety
Crash causation analysis
Large language models
In-depth accident investigation
Large language model
Fine-tuning
Journal
IF:
7.9
Papers:
4.7K
Citations:
3.2W
Organization
Cited Papers