Traffic accident data include multidimensional dynamic and static factors such as “people, vehicles, roads, and environment” at the time of the accident, which is one of the important data sources for improving the traffic safety environment. Based on the case data of traffic accidents and the construction idea of knowledge graph, the knowledge demand, knowledge modeling, knowledge extraction, and knowledge storage of traffic accidents are analyzed in detail. Finally, the traffic accident knowledge graph is constructed. The visualization analysis of accident is carried out from four angles: accident portrait, accident classification, accident statistics, and accident correlation path. The visualized graphic network displayed by the traffic accident knowledge combines human cognition with machine cognition, which improves human’s ability to understand massive and complicated data. The theoretical system of constructing traffic accident knowledge graph has certain reference significance in the follow-up research on the analysis of massive traffic accident data.