A Robust and Secure Framework for an AI-Assisted Academic Writing Platform
摘要
The Artificial Intelligence (AI)-assisted academic writing platform is a digital tool that leverages the use of artificial intelligence (AI) technology to help the platform users, i.e., students, teachers, and professionals in various aspects of academic writing. This work reviews existing AI-assisted academic writing platforms, discusses core system design features that are important considerations in designing such frameworks, and provides a critical discussion of the strengths and weaknesses of existing platforms. This work also proposes a robust and secure framework for an AI-assisted academic writing platform based on Azure cloud services. The platform aims to provide comprehensive supervisory support, free from time and location constraints, to help its users e.g., teachers in terms of helping with their workload in assessing academic reports, and to students by providing support to enhance their academic writing. The proposed framework is based on Microsoft (MS) Azure cloud services because it provides reliability, scalability, and a wide range of services such as Web Server and Content Management System (CMS), Azure Database for MySQL and OpenAI Azure Service i.e., ChatGPT in a secure environment. In addition, the proposed framework integrates with a Private Cloud to further enhance the platform’s functionality and secure environment. The main objective of the platform is to deliver an interactive system that provides reliable feedback on English writing in a range of academic fields. The proposed platform is then evaluated by piloting on the PolyU SPEED academic course led by 16 teachers and tested on academic reports of 273 students. The survey-based evaluation results show that the proposed AI-assisted academic writing platform is easy to use and showcases the ability to identify issues related to specific genre-based content and coherence, provide assistance in grammar and vocabulary, and highlight issues in reference style with reasonably high accuracy.