Determining Experts’ Neutral Weights Based on Cross Entropy-Like and Application in Evaluation with Incomplete Judgment Information

In order to mitigate the impact of systematic errors on evaluation results, the T-score method (standard score method) is widely employed at present. However, the underlying assumption for using this method is that the level distribution of contest entries reviewed by each expert remains consistent, which becomes challenging to achieve in evaluations with incomplete judgment information and consequently introduces significant errors in the evaluation outcomes. This study aims to address these limitations of the T-score method by introducing a weighting approach to enhance judging accuracy. To accomplish this objective, we introduce a novel concept called cross entropy-like based on cross entropy and propose a methodology for determining neutral weights of experts based on cross entropy-like. The reliability and scientific rigor of our proposed approach are validated through simulation experiments and comparative analysis against alternative methods. This study has important implications for the fair evaluations with incomplete judgment information.