Study on Text Classification of MOOC Course Comments Based on Chinese Character-level Convolutional Networks

Abstract

With the rise of the educational big data and the development of MOOCs, a wide range of comments have come up. Now most of the existing MOOCs text classification models are for small scale comments, and fail to extract high-level abstract information among characters. In this paper, crawler technology is used to crawl the comment data in MOOCs forums comprehensively, and a character table containing 5500 comment data is constructed then. In terms of word embedding operation, a text classification model of MOOCs comment data based on Chinese character-level convolutional neural network is proposed to avoid the pre-training process of word vector. The experiments show that our model could improve the classification accuracy comparing with traditional methods.