Towards Auto-Coding of Collaborative Interaction Texts Based on Maximum Entropy Approach

Content analysis is an important method in the research into collaborative learning. The usual method for deeper content analysis is to first categorize and code interactive texts and then analyze them. However, the content analysis method with categorizing and coding is time consuming in the first period of research for coding content analysis, which restricts greatly the amount and scale of the analyzed contents. Therefore, this paper proposes a method to automatically code the collaborative interaction text based on the maximum entropy approach. Taken a great number of interactive notes as input corpus, an experiment is conducted to test the efficacy of the proposed approach. The experiment results show that the coding effect of the proposed auto-coding approach is close to that of the sentence-opener approach in which the coding category is manually chosen by students. Additionally, as corpus increase and features are improved, the coding effects will be further strengthened.