| Summary: | It is of great significance to develop a scientific and reasonable evaluation system for improving the teaching quality in colleges or universities. Traditional teaching evaluation systems are always limited to subjectivity and injustice for single dimensional evaluation index, subjective weight setting, and inefficient integration. In this paper, neural networks are introduced to evaluate teaching quality in an objective way. Firstly, three sub-networks S-NET, P-NET and L-NET are trained on the evaluation sets of students, peers, and leaders. To boost the performance of a single network, we offer an integrating way to synthesize the results of three sub-networks. The integrated network I-NET is trained on the multi-dimensional data sets to acquaint comprehensive evaluations. The experimental results demonstrate the feasibility of three sub-networks with the results that the average error rate of S-NET is 1.457%, P-NET is 1.003% and L-NET is 1.528%. To be further, the prominent evaluation ability of I-NET is also verified with the accuracy of 98.59% on the verification data sets. Finally, online teaching evaluating system based on the SSM framework is designed to provide a convenient and friendly interface.
|
|---|