Random inspection evaluation of Shanghai graduate dissertation based on Bayesian decision tree

In order to improve the quality of graduate dissertation and examine the quality of graduate education, the mechanism of graduate dissertation random inspection evaluation in Shanghai has been operated for more than ten years. Evaluation experts evaluate the quality of dissertation by using subitem evaluation method rather than comprehensive evaluation method to reduce the risk of misjudgment. Decision tree is a model to explain the data processing from subitem evaluation to comprehensive evaluation. To reduce the disadvantage of decision tree algorithm, the Bayesian decision tree is applied for denoising. The experiment result shows that our method can predict the comprehensive evaluations effectively according to subitem evaluations.