One of the reasons why a project fails is that the project manager cannot correctly determine the skill level of the people in charge of tasks, and as a result, the task is not correctly assigned. In this research, we studied an inference system that determines the validity of task allocation based on a user model of the person in which system experience and skills acquired in the past are accumulated. The inference system determines from the degree of system similarity whether the skill that the person acquired in the past can be used for this task. In this paper, we propose a method for calculating the degree of similarity between the current system and systems experienced in the past based on a set system attributes for each skill. In addition, we developed a prototype system based on this model, evaluated it by using data from an actual development project, and confirmed its effectiveness.