A crash narrative is the detailed description of a crash by law enforcement officers. Although it is a common practice to manually review crash reports to collect valuable information from crash diagram and narrative, the process is time-consuming and labor intensive; and the interpretation of a crash depends on a reviewer’s experience, knowledge and judgment. This project will develop a method for extracting, classifying and analyzing the key contextual information and sequence of events relating to a crash using the text mining techniques and tools. Such information can be vital for seeking the causal factors of a crash such as the actions, behaviors and circumstances leading to a pedestrian crash that are not usually presented in the structured data fields. In addition, maximizing the value of a crash narrative will encourage and incentivize law enforcement agencies to use a narrative to capture things that are not available in the data fields.
Project Documents
Project Details
Project ID
MIL115784
Status
Complete
Start Date
October 1, 2019
End Date
September 30, 2021
Focus Areas
Data Analytics, Modeling and Simulation
Safety
Sponsors
Wi Dept Of Transportation
National Highway Traffic Safety Administration (NHTSA)
Research Centers
Institute for Physical Infrastructure and Transportation (IPIT)
Principal Investigator
Qin, Xiao
Director, Institute for Physical Infrastructure and Transportation (IPIT)
Professor, Civil and Environmental Engineering University of Wisconsin-Milwaukee
Co-Principal Investigator
Kate, Rohit J
Associate Professor Department of Computer Science University of Wisconsin-Milwaukee