基于分辨函数的科研项目评估专家知识获取研究

Study on Expert Knowledge Extraction of Evaluation for Scientific Research Project Based on Discernible Function

为了解决省级科研项目评估中专家知识难以获取的问题,提出了一种基于分辨函数的知识获取方法。该方法以粗糙集理论为基础,对预选评审指标,采集专家意见,构建决策表,生成分辨函数求解核和约简,从中发现隐含规则,提取专家共识,构建知识库。以甘肃省自然科学研究基金项目评估进行了仿真案例计算,计算表明该方法能够客观反映专家意见,适用于立项评估、中期评估、绩效评估等科研项目管理全过程,也可应用于其它科研项目评估决策中。

To solve the expert knowledge extraction problem in provincial scientific research project evaluation, this paper pro- posed a knowledge extraction method based on discernible function. This method is based on rough set theory. According to the pre- liminary evaluation indicators, it gathers expert opinion, sets up decision table, generates discernible function so as to get core and re- duced attribute set. From the reduced attribute set, It can find out the rules implied in decision table, extract expert consensuses and construct knowledge base. Taking Gansu Province Natural Science Research Foundation project evaluation as example this paper carried out a simulation calculation. The calculation shows that the method can reflect the opinions of the experts. It is suitable for project evaluation, interim evaluation, performance evaluation of scientific research project management in the whole process. It can also be applied to other scientific research project evaluation decisions.