Identifying traffic accident concentration area is important for road safety improvements. Previous spatial concentration detection methods did not consider the severity levels of accidents, and the final traffic accident risk map for the whole study area ignores the different users’ requirements.
This thesis proposes an ontology-based traffic accident risk mapping framework. In the framework, the ontology represents the domain knowledge related to the traffic accidents and supports the data retrieval based on users' requirements. A new spatial clustering method, called DBCTAR (Density-based Clustering for Traffic Accident Risk), takes into account the numbers and severity levels of accidents is proposed for risk mapping. To demonstrate the framework and the new algorithm, the Ontology-based Traffic Accident Risk Mapping (ONTO_TARM) system and a web-based clustering service GeoClustering have been developed. Four case studies in the city of Calgary with final risk maps are presented and discussed.