Journal of Tongji University(Natural Science), Volume. 53, Issue 7, 1074(2025)
Analysis of Influencing Factors and Severity of Chain Rear-End Collision Using Text Mining
Based on nearly eight years of data collected from Weibo, eight topics related to chain rear-end collisions were identified using the latent Dirichlet allocation (LDA) topic model and social network analysis, revealing the main characteristics and mechanisms of these accidents. The study finds that highways are the primary accident scenes. Accident frequency during rain and snow is higher than in foggy conditions. Failure to maintain a safe following distance and speeding significantly increase accident risk. By assigning topics to each document using LDA and extracting severity information through regular expressions, an ordered logit regression model was constructed to analyze the impact of different topics on accident severity. The results show that the average number of injuries and fatalities in chain rear-end collisions are 2.12 and 1.85 times higher the averages for general motor vehicle traffic accidents, respectively. The severity of chain rear-end collisions on highways is higher than at intersections, with an odds ratio (OR) 3.3 times that of intersections. The OR for foggy-related accidents is 9.4 times that of rainy and snow, while the OR for accidents involving trucks is 4.6 times that of cars is 2.2 times that of buses.
Get Citation
Copy Citation Text
WANG Ling, LI Yidan, WANG Zijian, ZHANG Long, XING Yingying, MA Wanjing. Analysis of Influencing Factors and Severity of Chain Rear-End Collision Using Text Mining[J]. Journal of Tongji University(Natural Science), 2025, 53(7): 1074
Received: Jun. 24, 2024
Accepted: Aug. 26, 2025
Published Online: Aug. 26, 2025
The Author Email: XING Yingying (yingying199004@tongji.edu.cn)