Work

AI-Assisted Academic Text Refinement and Integrity Analysis for Ethical Scholarly Writing

Year: 2026
Type: preprint
Abstract: We propose an AI-Assisted Academic Text Refinement and Integrity Analysis System designed to address the growing challenges of AI-generated content and plagiarism in scholarly writing. The system integrates natural language processing techniques to preprocess input texts, followed by a multi-faceted analysis that evaluates AI-likeness through linguistic features such as perplexity, lexical richness, and syntactic patterns. Furthermore, it employs web-based plagiarism detection using TF-IDF and cosine similarity to identify unoriginal content, thereby ensuring textual authenticity. If AI-generated or mixed-origin text is detected, the system applies advanced refinement techniques, including paraphrasing and tone enhancement, while preserving semantic integrity through validation using pre-trained language models. The proposed method also incorporates user management and submission tracking, enabling researchers to monitor their manuscript’s evolution and receive detailed reports with visualizations of AI-detection scores, plagiarism percentages, and readability improvements. Our contributions lie in the holistic integration of AI-detection, plagiarism screening, and humanization modules within a unified framework, which not only enhances ethical manuscript preparation but also provides actionable insights for academic writers. The system’s significance is underscored by its potential to foster academic integrity while adapting to the increasing prevalence of AI-assisted writing tools.
Language: English
Subfield: Education
Open Access status: green
OpenAlex ID: W7236136414
Cites: 29
Cited by:
Related to: 10