Abstract: In this paper, an adaptive evaluation model of the optical transport network (OTN) knowledge points based on the knowledge graph is constructed, which is used for the intelligent evaluation of the massive OTN knowledge mastery. The model combines the item response theory and the Bayesian theory to achieve an adaptive selection of questions in the process of OTN knowledge tests. Meanwhile, the model achieves a rapid calculation of the OTN knowledge level with the aid of the knowledge point relationship in the knowledge graph. It is demonstrated that the model can calculate the ability of those tested in real time and automatically adjust the difficulty of the next question based on the ability calculation results. Finally, the knowledge weakness of those tested is accurately predicted, with the prediction error less than 4%. Practical cases are used in this work to show the good application effect of the proposed model.