An Intelligent Word Processing Add-in for Real-Time Academic Argument Structure and Tone Enhancement Using NLP and LLMs

Students and academics find writing academic papers challenging, especially in terms of structuring arguments and adhering to a formal tone in their documents. This leads to suboptimal submissions and quality of writing. This paper describes the development of an AI-based Microsoft Word Add-in to help academic writers in real-time by evaluating their argument and detecting informal words and phrases within their documents. The Writing Assistant uses a Python FastAPI backend with OpenAI API for argument reasoning, and spaCy for NLP processing of sentences, while the user interface is implemented with React and Office.js for a floating sidebar design. It provides feedback on unsupported arguments by identifying absent Claims, Evidence or Reasoning, and suggests formal alternatives to informal language. Leveraging findings from related research, we would expect an LLM-based argument analysis to have an F1 score of 0.85 on argumentative datasets, and a transformer-based tone scoring to have a Pearson correlation of 0.81 with human ratings by experts. The add-in offers both features in a privacy-preserving stateless system without a database, distinguishing it from existing solutions such as Grammarly, ProWritingAid, and MS Editor. The tool aims for a response latency of less than five seconds and can be used by multiple users, making it ideal for academic institutes, as well as individual researchers. This work was conducted at Arab International University (AIU), Syria. The official website of the university is: https://www.aiu.edu.sy