一种基于Web2．0的协同评价与群体决策方法  

A Collaboration Evaluation Method and Group Decision Method Based on Web2.0

作　　者：余力[1] 黄云 高凯 张燕磊 

机构地区：[1]中国人民大学信息学院,北京100872 [2]Northwestern University, Evanston USA 60202

出　　处：《系统工程》2017年第6期145-151,共7页Systems Engineering

基　　金：国家自然科学基金(No.71271209,No.71331007);中央高校基本科研业务费专项资金等资助

摘　　要：评价与群决策是管理科学中重要问题。传统群体评价决策中通常赋予专家相同的评价权重,但实践中,可能存在少数专家评价时的主观故意舞弊行为,或因客观上业务水平的限制等原因,导致其做出不准确评判,甚至是极端评价,影响最终评价结果。本研究把Web2.0协同的思想应用到评价中,提出协同评价理论,并在此基础上提出一种具体的群决策专家评价权重的协同分配方法,可自动赋予评价专家不同的评价权重,并能有效自动检测奇异评价分。实验表明,在出现少数极端评价的情况下,协同评价方法仍具较强的稳定性。Evaluation and Group Decision is one of key issues in management science. For traditional group decision methods, all evaluators are assigned same weight. Unfortunately, it often exists that irregular individual ratings deliberately are given by a few evaluators because of their good （or bad） private relationship with the evaluated persons, or unfair rating are given by a few evaluators who are hard to give fair rating because of their limited professional abilities. In this paper, inspired by the collaboration idea from Web2.0, collaborative evaluation theory is firstly proposed, a specific collabo- rative evaluation methods are presented. By these methods irregular rating can be automatically detected and each evalua- tor is assigned a different evaluating weight according to his rating accuracy which is collaboratively decided. Penalty factor is proposed to penalize the irregular rating. A real application for course presentation evaluation is implemented and effect of penalty factor is checked. The experiments show that the proposed methods achieved stable evaluation results and better advantage over traditional methods even though irregular evaluations are made.

关 键 词：协同评价 群决策 评价方法 WEB2.0 

分 类 号：N945[自然科学总论—系统科学]