With English as the dominant global language, demand for precise grammar processing grows rapidly. Traditional methods struggle with complex grammatical structures, especially for non-native speakers. To address this, we propose a deep neural network (DNN) and transfer learning-based system for automated English grammar parsing and correction. Our approach: (1) Uses Transformer-based DNNs to capture intricate grammar patterns from large corpora; (2) Applies transfer learning to adapt generic linguistic knowledge to specialized error correction; (3) Achieves 10% higher accuracy and 8% better recall than traditional tools. Critically, it handles complex errors like subject-verb disagreement and tense misuse with high robustness.
Liu et al. (Mon,) studied this question.