Abstract:The evaluation of scientific research project is a multi-attribute group decision-making problem, which is determined by a variety of interactive indicators. And the relationship between indicators is not independent, they constitute a complex gray system with incomplete information. Based on the fuzzy theory, a fuzzy semantic table is proposed, which is filled in by percentage, and the fuzzy semantic algorithm is used to build the project evaluation model of scientific research projects. In the construction process, the electronic scoring table is recovered first, and the effective value of the fuzzy value greater than the fuzzy median is taken, and the main index and sub-index of the evaluation of scientific research projects are extracted, and the evaluation table of scientific research projects is established for the evaluation of the director, deputy director and the overall leader of the experts of the evaluation committee. The evaluation experts give the weight of each criterion according to the evaluation table of scientific research projects, and grade the scientific research projects according to this criterion. The fuzzy set method is used to express, and each abstract criterion is transformed into a quantitative triangular fuzzy number to evaluate the ranking of the fuzzy numbers. Finally, the case evaluation and analysis are carried out to verify the model, the results show that it has good superiority and provides a more feasible method for scientific, objective and fair evaluation and optimization of scientific research projects.