Sentiment Analysis of Course Evaluation Comments in Chinese
Ke Li, Fei Hu · 2024
A wealth of comments that students provided with online course evaluations can be valuable in helping improve and refine the teaching. However, it is difficult for the computer to utilize these comments and understand student appeal because the comments, expressed in text format, cannot be effectively represented in a way that the computer can recognize and the meaning implied in the comments is always blurred that deepens the difficulty for the computer to understanding them. In this paper, we tackle the representation problem by projecting comments (including words) into a dense vector space, proposing to capture the meaning of the text by exploiting the Gated Recurrent Unit model (GRU) and a real-world dataset of five-year comments (during year 2013–2017) in a university's end-of-course evaluations for the Chinese language. The proposed model is evaluated on the ability to judge emotional tendencies of these comments, discovering useful suggestions and criticisms. Also, we improve the effect of judgement by putting GRU into a highly deep neuron networks and taking into account both the forward and backward association of words in the comment.