The evaluation problem in educational environments has been a constant, even today due to the current contingency generated by the COVID-19 pandemic, education has had to migrate to the virtual environment and the difficulty of evaluation persists. The purpose of this work is present comprehensively the efforts made in the construction of an intelligent system for the customization of evaluation activities implemented in virtual learning environments. The system architecture consists of three modules. In previous works, an architecture of a strategic learning meta-model has been designed that is divided into three layers, the intelligent layer where an ontological model was designed, the infrastructure layer has been designed with a cloud computing solution and the reactive layer that contains the architecture of the regulation model. Each of these layers corresponds to a module of the system. This work presents the results of three stages of experiments and shows a continuous increase in the approval rate trend.