Research on Automatic Scoring System of College English Writing Based on Natural Language Processing

Writing scoring in college English teaching brings a great workload to teachers. The development of an automatic writing scoring system based on natural language processing technology can effectively improve the scoring efficiency. The scoring system uses a deep learning model to build a semantic feature extraction framework, identifies language units through word segmentation module, and captures semantic information by combining word vector technology. The scoring engine designs a multi-dimensional evaluation system to realize quantitative evaluation of writing quality. A distributed architecture is adopted in system engineering to ensure the efficient and stable scoring service. The evaluation results show that the system's scoring accuracy is 82%, the scoring of a single composition takes 1.2 seconds, and the service availability is maintained at 96%. A practical application display system can effectively support writing teaching and provide students with timely evaluation feedback.