基于BP神经网络的高校院系科研绩效评价模型

Performance Evaluation of Scientific Research in Universities:Based on BP Neural Network

本文在分析当前各种高校院系的科研绩效评价指标体系的基础上,针对目前各种评价指标多、定量难、人为干扰因素多等问题,采用主成分法从大量评价因素中筛选出绩效评估的主要因素,即在保留评价信息的前提下对数据进行有效降维,并通过BP神经网络的自学习功能计算出高校院系的科研绩效。该模型能较为科学地评估高校院系科研团队的科研成效、核心竞争力及研究潜力,为高校科研管理提供一种新方法。

After analyzing various evaluation index system of scientific research performance in universities, problems such as too many indices, hard-quantification and easily-correlated indices, are studied in this paper. Principal Component Analysis is adopted firstly to screen out the main one from a great deal of evaluation factors to reduce the dimension effectively under the condition that evaluation information is reserved. Quantitative evaluation of scientific research performance in universities is gained by the self-study function of the BP neural network. This model can reasonably evaluate scientific research performance, key competition ability and research potentiality of scientific research teams in colleges/departments and provide approaches for scientific research management in universities.