加权朴素贝叶斯模型在高校学科评价中的应用

摘要

Discipline evaluation is an important part in higher education evaluation. It plays a significant role in discipline construction in colleges and universities. It is significant challenge how to adopt scientific discipline evaluation to choose advantageous disciplines and newly-emerging ones. This paper proposes a discipline classification method based on Weighted Naive Bayes model. It establishes Discipline Classifier according to weighted discipline evaluation attributes and mutual information. An automatic discipline classification system is implemented, verifying the method and analyzing data from universities in Shanghai. Experimental results show that the used scheme can achieve about 86.67% accuracy in forecasts. It provides advice and guidance for discipline evaluation, and establishes foundation for discipline development strategy.%学科评价是高等教育评估的重要内容，是加强高校学科建设的重要手段。如何采用科学的方法开展学科评价，客观地对学科进行分类，遴选优势学科和潜力学科在学科建设中至关重要。提出了一种基于加权朴素贝叶斯模型的高校学科分类方法，依据不同权重的学科评价属性和互信息量计算综合权重，构造高校学科分类器，实现了一个自动化的高校学科分类预测系统，并对上海高校学科样本进行应用分析。实验结果表明，应用基于加权朴素贝叶斯模型的高校学科分类预测系统对学科进行预测分类，准确率达到86.67%，为高校学科综合评价提供了基本依据，为新一轮学科发展战略的制定奠定了基础。