A Topic Analysis Model for Chinese students' compositions

摘要

This paper proposes an innovative Topic Analysis Model and describes its application to efficiently analyze the contents of Chinese students' compositions. With Topic Analysis Model, we can analyze composition's semantic content related to latent topic information. First, we judge whether the tested compositions are on-topic or not, then give the highly probable words reflecting the topic of the tested compositions and the sample composition set respectively. Our testing experiments on actual data set shows that the Topic Analysis Model has a consistently good performance in semantic content analysis on topics for Chinese students' compositions, so the resulting Topic Analysis Model can give better predictive performance and uncover interesting statistical information in terms of topic analysis for the Chinese students' compositions.