The Wisconsin Department of Transportation awarded $75,000 to Xiao Qin for a 10-month project that aims to improve the state’s highway safety by further improving data analysis related to crashes. Qin is a Lawrence E. Sivak ’71 professor of civil & environmental engineering in UWM’s College of Engineering & Applied Science and director of the university’s Institute for Physical Infrastructure and Transportation.
“Our goal is to identify the most common patterns in crash sequences,” Qin said.
To do this, Md Abu Sayed (PhD Civil & Environmental Engineering ‘22) and Qin will develop an algorithm that automatically extracts data from crash narratives. These narratives are written by law enforcement officers and detail the sequence of driver actions and vehicle movements preceding, during and following a crash. They often include information related to human behavior, which contributes to between 80 and 90 percent of crashes.
“The details found in crash narratives shed light on opportunities for effective safety interventions,” Qin said. The problem, he said, is that such data is not easily captured.
Work builds off nationally-funded, $100K project on crash narratives
The work builds upon Qin’s $100,000 project with WisDOT in 2021, research that was funded through the National Highway Safety Traffic Administration. In that project, Qin and Sayed used technologies (including machine learning, natural language processing and feature extraction) to capture and analyze valuable information from unstructured data, such as crash narratives, that relate to human behavior.