An Algorithmic Study on Improving the Quality of English Writing Teaching to College Students in an Intelligent Environment

Abstract

In response to the need to improve the quality of English writing instruction in higher education institutions in an intelligent environment, this paper proposes a writing diagnostic system based on Jieba word segmentation and the EVCM model. A teaching model framework oriented toward interactive instruction is constructed, and the Jieba word segmentation algorithm is optimized for text preprocessing. The EVCM model is designed to achieve lexical coherence diagnosis from three aspects: entity distribution, coreference resolution, and conjunction detection. The stability and diagnostic accuracy of the system under different network environments are verified through simulation experiments. Teaching practice activities were conducted, and T-tests and Pearson correlation analysis were used to explore the impact pathways of the system on improving the quality of English writing instruction. Teaching experiments showed that the pre-test mean score for the high-performance group was 8.95, which increased to 9.96 in the post-test. The mean score differences between the pre-test and post-test for the medium- and low-performance groups were both greater than 1.5 (p < 0.001). Students grouped by English writing performance exhibited significant differences in the overall use of the English writing diagnostic system and in four dimensions: self-motivation, monitoring and diagnosis, execution, and result generation. The use of the system not only directly affects students' writing proficiency but also indirectly promotes ability development by enhancing their writing motivation and beliefs.