Scientific Text Fingerprinting in the Age of Artificial Intelligence

Abstract

Plagiarism detection systems have become a critical component in safeguarding academic integrity, particularly in light of the rapid expansion of digital content and the emergence of text generation tools based on large language models. This study provides a systematic and comprehensive examination of plagiarism detection techniques grounded in scientific text fingerprinting within an artificial intelligence framework, highlighting the methodological shift from literal string matching to semantic detection of paraphrasing, translation, and obfuscated plagiarism. The article adopts a systematic literature review approach and proposes a multi-layered hybrid framework that integrates text fingerprinting and local algorithms, semantic text representations through embedding’s, and language-specific processing layers addressing the morphological and syntactic challenges of Arabic. Furthermore, the study discusses standard evaluation protocols and benchmarking metrics, and offers a critical analysis of the strengths and limitations of existing approaches, alongside practical recommendations for academic institutions and scholarly publishers.