摘要
对群决策及残缺主观评价系统研究分析,发现直接对评分取均值排名会因专家打分尺度的不同带来较大误差,虽然T分数法可以将打分尺度统一,但是缺乏对专家评价对象不同的考虑.为减小此系统误差,提出了一种确定专家权重的数量积法,并将其应用于论文型竞赛排名系统.与传统方法及T分数法比较,对T分数求加权平均值排名,提高了主观名次与客观名次的重合度,减小了其乱序度.计算欧氏距离,证明新方法减小了评委打分的误差度及论文的争议度.
In the analysis of group decision making and incomplete subjective evaluation system,we find that taking the means ranking directly brings big error due to the difference of experts＇ scoring scales. Although the T-score method can be used to unify them,it lacks the difference of experts＇ evaluation objects. In order to reduce this system error,in this paper,we propose the Dot Product method for determining experts＇ weights and apply it to the theses type competition ranking system. Compared with the traditional and T-score methods,this method improves the coincidence of the subjective and objective ranking,and reduces the disorder degree. By calculating the Euclidean distance,we prove that the new method can reduce the experts＇ scoring error and the controversy of theses.
出处
《烟台大学学报（自然科学与工程版）》
CAS
2015年第4期249-254,共6页
Journal of Yantai University(Natural Science and Engineering Edition)
基金
中央高校业务费资助项目(B12020128)
关键词
主观评价
数量积
权重
残缺评分矩阵
系统误差
subjective evaluation
Dot Product
weight
incomplete score matrix
system error
