The safety of elevator, as a closely related equipment in people's daily life, is getting more and more attention.Once an elevator accident occurs, it will bring strong social impact.It is important to control the elevator safety risk from the source to reduce the accidents.This article combines the risk factor analysis methods of Knowledge Graph (KG), Decision Making Experimentation and Evaluation Laboratory (DEMATEL), Interpretative Structural Modeling (ISM) method, and Multiplication of Cross Matrices (MICMAC) to analyses elevator safety accidents in a data-driven manner in order to minimize the reliance on experts.Firstly, the causal network of accident occurrence is extracted from the accident reports based on the knowledge graph approach.Secondly, complex correlations in the causal network are quantified from the application of DEMATEL.Then, the ISM method was used to construct a hierarchical structure of elevator safety risk factors, on the basis of which the MICMAC method was combined to classify the types of risk factors and analyses the degree of influence of risk factors at each level.Finally, targeted preventive measures for elevator safety accidents are proposed based on the results of the model analysis.