A Framework for Detecting and Summarizing Students' Typical Errors in English Teaching

Abstract

With the development of online education, the mining and analysis of educational data has become especially important. In teaching, detecting students' typical errors is an extremely important factor for higher teaching efficacy. Most of the current researches use clustering or decision tree algorithms for partitioning. However, these algorithms more or less fail to recognize the connection between students and the errors they make, and cannot effectively and intuitively derive their typical errors. This paper proposes a framework that combines community detection and association rules to detect students' typical errors in English teaching. First, the framework adds the error auxiliary nodes and obtains the student's error communities and typical errors. Second, it calculates the errors' frequent itemsets, and mines the association rules between errors. And last, it combines the association rules with the error communities to supplement the potential errors, which effectively summarizes students' typical errors in their learning process.