Abstract
The aim of this study is to construct the traffic accidents authentication models for the Local Traffic Accident Authentication Committees (LTAAC) as reference. In present, due to the committee members will change every two or four years and different LTAACs will authenticate the similar case with different results. This study will build a database which includes 5,268 client data, and the collision types include car/car, car/motorcycle, and motorcycle/motorcycle. This study utilizes the artificial neural network method (ANN), and the classification tree method (CT) to construct the models. This study shows that both of authentication models all have over 70 percent accuracy in the accident responsibilities.
