Neuro-Symbolic Architectures for Explainable Multi-Modal Plagiarism Detection in Academic Assessment
Paul Showemimo · International Journal of Advances in Engineering and Management · 2025
The pervasive integration of Artificial Intelligence (AI) into educational ecosystems has introduced both unprecedented opportunities and complex challenges, particularly concerning academic integrity. While AI-driven tools have become indispensable for identifying traditional forms of plagiarism, the emergence of sophisticated content generation models and multi-modal assignments necessitates a paradigm shift in detection methodologies. This paper proposes a novel MultiModal Neuro-Symbolic Assessment System (MNSAS) architecture specifically designed for explainable plagiarism detection. Our approach integrates the robust pattern recognition capabilities of neural networks with the logical reasoning and transparency of symbolic AI to process and analyze diverse academic submissions, including text, code, images, audio, and video. This neuro-symbolic synergy not only enhances the accuracy and robustness against advanced obfuscation techniques but, crucially, provides humanunderstandable explanations for detected anomalies, fostering trust and promoting genuine learning. We detail the technical components, architectural considerations, mathematical formulations, evaluation metrics, and research questions that underpin this innovative framework, demonstrating its potential to significantly advance the field of academic integrity and contribute to a more transparent and equitable educational landscape.