Optimization model for fair judging of large-scale innovation contests based on experts’ neutral weights

Abstract

The group decision-making method and standard score method are commonly used for evaluating large-scale innovative contest entries. However, the standard score method assumes that the quality distribution of entries evaluated by each expert remains consistent, which is difficult to achieve in practice due to different experts reviewing different entries. To enhance fairness in judging results for such contests, we propose an improved standard score method based on weighting. First, we establish a mathematical model for cross-assignment of entries and design an algorithm to solve it while discussing one property on its optimal solution. This model enables any two experts to be assigned as many identical entries as possible, facilitating reliable pairwise comparisons matrix while avoiding assigning the best or worst entries to multiple experts in one judgment. Second, inspired by cross entropy concepts, we introduce a definition of “cross entropy-like” and establish a mathematical model using this concept along with minimum sum of squared errors method to determine neutral weights for experts. Finally, experiments demonstrate that our proposed improved standard score method is more scientific and reasonable than traditional standard score method while enhancing objectivity and fairness of evaluation results.