Research on an English Essay Evaluation and Personalized Suggestion System Based on BERT and Artificial Intelligence Models

Abstract

Artificial intelligence technology is now widely used in the world, among which BERT (Bidirectional Encoder Representations from Transformers) model in the field of natural language processing has received much attention. Because BERT model shows powerful functions and effects in such aspects as essay grading, grammar correction, vocabulary upgrading and so on. In order to give students a comprehensive and comprehensive feedback on their English writing, and let them improve their grammar, vocabulary and structure according to the feedback, our paper proposes a BERT scoring model embedded in a system for English essay evaluation and personalized suggestions generation. In our system, the first step is to input students' English essays and score them in BERT model. The second step is to optimize and upgrade grammar, vocabulary and structure according to a complete deep learning model based on students' English essays and scores, and provide suggestions to students. For example, in terms of grammar, the model will correct students' mistakes and help them reach the correct level in grammar; in terms of structure, the model will assign some targeted suggestions to students in terms of long sentences and paragraphs; in terms of vocabulary, the model will give students some suggestions for replacing common words with advanced words. The third step is to input all the above feedbacks into the Qwen dialogue model and have one-to-one interaction with students to provide one-to-one dialogue so that students can know what they need to modify and how to modify it. Our system is equivalent to an intelligent, interactive, one-to-one personalized robot English composition teacher.