Aiming at the shortcomings of traditional system evaluation methods that rely too much on the subjective opinions in the construction of an evaluation indicator system and weight distribution, this article utilizes the similarity between the structure of analytic network process(ANP) and the structure of knowledge graph to proposes a strengthened analytic network process method based on knowledge graph. This method adopts prior knowledge to enhance the completeness and objectivity of the evaluation. Firstly, the text data related to the evaluation task is collected to establish the knowledge graph which is the indicator library of the network layer. Secondly, the control layer of the analytic network process that includes the evaluation objective and the evaluation criteria is determined according to the evaluation task. The network layer is constructed by searching for suitable indicators in the network layer indicator library according to the control layer. Thirdly, the similarity between the indicators and the criteria is adopted as the objective standard to adjust the relative importance of the indicators. Through the analytic network process, the weight of each indicator is calculated to complete the system evaluation based on the scrore of each indicator. Finally, the results of the application cases verify that the proposed method has effectiveness, advanced nature and universality.