基于广义评价局部环境的多线群组评价方法研究

Classic evaluation methods exist problems such as the basic assumptions mislead the results, the methods doesn’t apply the group scale and the sensitivity problems of the methods. Now most of group evaluation methods design continue the classic ideas, and a certain scale group evaluation problem is still not solved. Using group decision-making, complex system and nonlinear programming methods, we design the multilinear group evaluation methods based on generalized partial evaluation environment.. Research: Discuss general group evaluation methods base on the partial evaluation environment which correspngding to group evaluation problem. Advanced research in group evaluation method based on both the group preferences having distribution and the structural characteristic analysis of the partial evaluation environment, excepting the comprehensive evaluation methods looking after the group consensus criteria. One conclusion is made in partial evaluation environment group evaluation model: the experts accommodate their estimations and the evaluation flow path via the best evaluation actions so that the group preferences and the utility could convergence in evaluation process. On the basis of above research , the project research in the following order: first, summarizes the structural characteristics of partial evaluation environment which in stable situations, combined with network mechanism model and social network analysis. Disscusion the integration of relationships (including relationships generate, change and superimpose etc.), the type classification of experts( mainly analysis the limited independence and limited legitimacy experts ) and optimization goals setting (satisfying both can be transformed and sequential rationality) respectively. Second, analysis about the basic building conditions that having long-lasting effective partial evaluation environment. The stochastic processes and complex network evolution model would be used. Furthermore, analysis the basic building conditions about general partial evalaution environment, which could envelopment natural characteristics and evaluaitional trends of the subject we studied. We will analysis the general evaluation environment stable situation instead of the sensitive problem of evaluation methods. There exists many evaluation processes which all convergence the partial evaluation environment stable situation .Also there exists multilinear evaluation methods could solve the problem based on partial evaluation environment. The methods above are used to reality. And the reality anaysis results are used to midify the models. . Significance: The conclusion could be used following: evaluation optimal goals setting numerical methods, normalize the evaluation rules setting methods and long-term evaluation mechanism theory just like innovation long lasting evaluation problem.

对经典评价方法应用中存在方法基本假设影响评价有效性、评价方法不适用较大规模的评价群体及评价方法敏感性等问题，综合运用群决策、复杂系统和非线性规划等理论与方法开发基于广义评价局部环境的多线群组评价方法。. 研究内容：基于评价基本要素（评价关系、评价者类型、评价目标设置和评价规则）讨论评价群体状态特征，结合网络机制模型研究，归纳稳定状态评价局部环境一般性特征，发现稳健评价关系。由评价目标、评价规则设置和评价要素分析可持续、可调节的长效评价机制设置，分析能够包络被评价对象基本特征和趋势性变化，混合负载评价的广义评价局部环境构建条件。以群体状态特征和演化规律，异构多渠道、多模态评价信息有效集结为重点，设计多线群组评价方法。. 研究意义：完善评价目标和评价规则设置与优化，提高评价方法适用性，为诸如创新可持续性评价等实际问题提供方法设计思路

评价群体连接状态直接影响评价方法效率和评价结论。具体地，评价局部环境稳定状态是具有停时特征的评价目标寻优问题，且多线评价过程收敛的评价方法存在效率问题有（1）具有时间特征的评价指标关于评价方法的选取截面敏感，并影响评价目标；（2）评价方法设计控制参数数量影响显著，即属性参数超过一定数量时评价网络很难涌现结果，意见收敛速度过慢；（3）评价方法参数设置不全、调参范围模糊，使得评价过程不连续，难以实现评价过程有序管理。研究发现预置群体先验偏好控制拐点影响大群体意见趋势。假设评价过程中大群体偏好分层，且随评价群体认知程度提高评价效率控制措施边际效果递减；评价者偏好为理性选择和价值诉求交替的个体评价互补策略优化，而评价者通过评价策略期望值回归估计其互补策略效用。面向大群体评价行为有序管理的需求背景，围绕评价行为、群组状态、评价过程引导3个阶段的关键技术，研究评价者个体策略及其局部交互规则，对比分析8个常用策略对评价过程和群体意见的影响，寻找个体策略和评价规则设置的可能阈值以保障评价方法效率。分析评价规则诱导的评价网络重新连接对评价方法效率的改进。具有自我赋权能力的预置群体自适应群组评价方法以群体意见动力模型为基础，考虑预置群体泛尺度动力驱动特征设计群组评价闭环控制方法和预置群体的自适应寻优控制评价方法。通过服务型制造网络Holon协同需求评价和知识员工绩效评价问题进行检验，结合评价方法应用需求构建评价目标分解模型。依托本项目研究，开发的评价方法已应用于相关单位技术干部绩效考核体系设计工作，目前已经过第二轮方法测试修正。同时相关评价方法在沈阳市皇姑区稽查和绩效考评办公室，黑龙江省绥化市北林区张维镇人民政府、绥化市供水总公司和黑龙江省林业设计研究院等单位进行推广和验证，目前已收回第2轮反馈问题和建议。本研究一定程度揭示了复杂群组评方法的设计机制，对提高方法使用效率有一定意义。