Exploring Innovative Strategies for Teaching English to College Students Based on Data Mining Theory

Abstract

This paper suggests a J2EE framework-based online intelligent diagnostic assessment scheme. The system has the capabilities of teaching, diagnosis, testing, and feedback as auxiliary teaching software through the automated alteration of subjective and objective questions and the individualized design of diagnostic results. Combined with the characteristics of English composition, the research focuses on the use of the shortest editing distance, cosine vector space, regular matching and other techniques to automate the composition correction and diagnosis, which not only realizes the final scoring of subjective questions but also the diagnostic report can provide students with personalized comments, precisely give the strengths and weaknesses in terms of knowledge points, as well as the analysis chart of knowledge points, scores and ability values of previous exams. In this paper, when evaluating the system with teacher ratings, the overall mistake rate for class essays was 9.0% when the similarity criterion S=0.2 was employed. For individuals and classes passing the test, the error was within 10%, and the score difference was within 2 points. Finally, the common errors of students in English writing were detected, with grammar errors accounting for 58.2%, vocabulary errors for 26.7%, and essay errors for 15.1%. Therefore, this system can make a good contribution to the teaching reform of English and achieve a real reduction of the burden for teachers and students.

Keywords

• english language teaching
• data mining
• shortest edit distance
• cosine vector space
• regular matching
• 97q70