AcademiCraft: Transforming Writing Assistance for English for Academic Purposes with Multi-Agent System Innovations

保存先:
書誌詳細
出版年:Information vol. 16, no. 4 (2025), p. 254
第一著者: Du Zhendong
その他の著者: Hashimoto Kenji
出版事項:
MDPI AG
主題:
オンライン・アクセス:Citation/Abstract
Full Text + Graphics
Full Text - PDF
タグ: タグ追加
タグなし, このレコードへの初めてのタグを付けませんか!
その他の書誌記述
抄録:In the realm of academic English writing, both native (L1) and non-native (L2) speakers face significant challenges due to the complex linguistic structures and conventions required. Existing writing assistance tools, while useful for grammar correction and text enhancement, often fall short by providing only corrected output without explanations, and they typically operate as opaque, proprietary systems. This study introduces AcademiCraft, a novel EAP writing assistance built upon a multi-agent system (MAS) and advanced large-language models (LLMs), integrating state-of-the-art natural language processing (NLP) and linguistic research. AcademiCraft distinguishes itself by correcting and refining academic texts and offering detailed explanations for its revisions, thereby deepening the user’s understanding of writing conventions. The tool’s open and replicable framework contrasts sharply with the closed nature of most commercial products. Through a series of rigorous scientific benchmarks, AcademiCraft has been shown to outperform leading commercial writing assistance tools in several key areas, including its ability to handle complex syntactic structures, improve coherence and cohesion in academic texts, and provide contextually accurate word choices, all while maintaining high levels of grammatical precision. This paper details the methodologies and underlying principles of AcademiCraft, and presents a comparative analysis of its performance, demonstrating its superior capability in supporting both L1 and L2 writers in mastering the complexities of academic English writing.
ISSN:2078-2489
DOI:10.3390/info16040254
ソース:Advanced Technologies & Aerospace Database