基于图模型和最优化的评价方法

发布时间：2018-05-24 18:31

本文选题：评价 + 图模型　；参考：《系统工程学报》2013年03期

【摘要】：针对评价问题中由于评价者的评价基准不同,难以给出一个符合原始评价信息的综合评价结果的问题,以及普遍存在的数据缺失的现象,建立了一个新的评价方法.运用了图论的相关原理,将评价对象视为点,评价者的评价行为将点与点连接起来构成边,由此得到评价的图模型,并转换为与之等价的二维邻接矩阵;运用最优化原理,将该矩阵向一维空间做投影,得到综合的评价结果.通过定义评价的合理性指标,并做模拟数据和真实数据的实证分析,验证了方法的合理性和实用性.结论表明:该方法充分反映了原始数据的排序信息,可以处理存在缺失数据的情形,并可用于大规模数据的评价问题.
[Abstract]:A new evaluation method is established to solve the problem that it is difficult to give a comprehensive evaluation result which accords with the original evaluation information and the phenomenon that the data is missing because of the different evaluation criteria of the evaluator in the evaluation problem. The evaluation object is regarded as a point by using the relevant principles of graph theory, and the evaluation behavior of the evaluator connects the points with the points to form an edge, and the graph model of the evaluation is obtained and converted into an equivalent two-dimensional adjacency matrix, and the optimization principle is used. The matrix is projected to one dimensional space and the comprehensive evaluation results are obtained. The rationality and practicability of the method are verified by defining the rationality index of the evaluation and making the empirical analysis of the simulation data and the real data. The conclusion shows that this method can fully reflect the sorting information of the original data, can deal with the situation of missing data, and can be used for the evaluation of large-scale data.
【作者单位】：哈尔滨工业大学管理学院;
【基金】：国家自然科学基金资助项目(71271070) 高等学校博士点专项基金资助项目(20092302110060) 教育部新世纪优秀人才支持资助项目(NCET-08-0171)
【分类号】：C931.1

【参考文献】