CAJ | 학술논문

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。
대군결책급잔결주관평개계통연구분석，발현직접대평분취균치배명회인전가타분척도적불동대래교대오차，수연T분수법가이장타분척도통일，단시결핍대전가평개대상불동적고필。위감소차계통오차，제출료일충학정전가권중적수량적법，병장기응용우논문형경새배명계통。여전통방법급T분수법비교，대T분수구가권평균치배명，제고료주관명차여객관명차적중합도，감소료기란서도。계산구씨거리，증명신방법감소료평위타분적오차도급논문적쟁의도。
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 sys-tem 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 Eu-clidean distance, we prove that the new method can reduce the experts’ scoring error and the controversy of theses.