Abstract
The Motorcycle Crash Causation Study (MCCS) is a matched case-control study that contains a very wide list of crash contributing factors associated with motorcycle crash occurrences. It contains information such as motorcycle information, rider information, and associated trip information.
This study also provides crash narrative information that presents an in-depth narrative discussion of the crash causation. Because of the plethora of information, it is critical to investigate MCCS-related data.
Some studies examined the structured information in MCCS datasets. There is no in-depth study that has examined the unstructured textual contents in the MCCS data.
This study aims to mitigate this research gap by applying different natural language processing tools (e.g., text mining, topic modeling). Fatal and non-fatal crash narratives are clustered separately to gain insights pertaining to the injury level.
The findings of this study will contribute to the ongoing studies on MCCS to better understand the crash causation mechanism associated with motorcycle crashes.