LAUSR

The evaluation of scientific research projects is a multi-expert decision-making problem under incomplete information environment. Whether evaluation results are reasonable or not depends on the expression of experts’ opinions, the reliability of experts and the aggregation method of experts’ opinions. From the perspective of expert reliability, this paper proposes a data-driven evidential reasoning method based on two-dimensional frames of discernment. In the proposed method, project evaluation information and experts’ characteristics information are used as two-dimensional evidence to represent decision information and decision quality information respectively, and belief distribution is used to express these two kinds of information. Experts’ characteristics information is used to measure the reliability of experts to modify project evaluation information given by experts. The discounted project evaluation information is aggregated by using evidential reasoning analysis algorithm to complete the evaluation of scientific research projects. In addition, a learning optimisation model is constructed to determine the relevant parameter values of the proposed method in a data-driven manner using historical evaluation data of projects. The empirical analysis of the National Nature Science Foundation of China verifies the validity and applicability of the proposed method.