Network Algorithms for Intelligent Evaluation of Composition in Middle School English Cloud Classrooms

Abstract

This study combines link grammar (LG) detector with N-grammar model to analyze and evaluate grammar in compositions. And then the composition level is judged through information entropy. Finally, the composition score is calculated based on the overall composition level and grammar weight. The experimental results show that the combined weight of recall and accuracy of the proposed method in this study is 89.9%, which is 26.6% higher than LG and 9.7% higher than Grammarly. In the performance test of scoring the entire essay, the proportion of error between the proposed method and manual evaluation is 87.29%, with a lower overall mean square error of only 3.08 and a shorter average running time of only 22.69 seconds. The method proposed in this study has a high accuracy and strong applicability in the evaluation of English compositions for middle school students, providing a new approach for teaching English writing for middle school students.