Contents I. Introduction
As the primary medium of communication in the global academic and professional realms, the accuracy and fluency of English writing are crucial for effectively conveying ideas. However, despite significant advancements in technology within the realm of English error correction, existing systems still face numerous challenges when handling coherent text. Most current error correction systems focus on grammatical analysis and correction at the sentence level, performing well with standalone sentences; however, they often struggle when faced with paragraphs or entire documents. [1] The sentences within a text are semantically interconnected, forming an organic whole. The limitations of existing error correction systems lie in their difficulty in capturing the logical relationships and contextual information between these sentences. This shortcoming can lead to correction results that deviate from the original intent of the text, and may even introduce new grammatical or semantic errors, thereby affecting the effectiveness of communication. For instance, consistency in tense, clarity in pronoun reference, and appropriate use of conjunctions all need to be considered within a broader context. Without a thorough understanding of the surrounding context, grammatical modifications based solely on individual sentences are unlikely to ensure the overall coherence and logic of the text.
