• Soumitra Jadhav
• Kalyani Dinde
• Shraddha Kharatmal
• Pradnya Jawale
• Tarun

The rapid digitization of academic and professional sectors has necessitated automated systems for assessing textual data, yet widespread adoption is often hindered by the "black-box" nature of traditional Artificial Intelligence (AI) models. This paper proposes a Universal AI Grader, a novel framework designed for the automated and transparent assessment of diverse content types, including essays, professional resumes, and source code snippets. Developed using Python and Streamlit, the system integrates a multi-modal text analysis pipeline capable of parsing complex formats such as PDF, DOCX, and images through Optical Character Recognition (OCR). The core architecture employs specialized Tf-idf and Logistic Regression classification pipelines tailored to domain-specific rubrics. A critical innovation of this work is the end-to-end integration of Explainable AI (XAI) techniques—specifically Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP)—to provide granular transparency into grading decisions. Experimental results demonstrate the system's ability to categorize documents accurately while offering visual, feature-level justifications that highlight the specific tokens influencing the final grade. By bridging the gap between high-performance automation and human-interpretable logic, this research fosters greater trust in AI-driven educational and recruitment assessments.

Explainable AI (XAI) , Automated Grading System , Machine Learning , Natural Language Processing (NLP) , LIME , SHAP , Multi-Modal Text Analysis , Optical Character Recognition (OCR) , Streamlit , Model Transparency

"XAI-Enabled Multi-Modal Text Analysis for Automated and Transparent Grading Systems", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 1, page no.b221-b224, January-2026, Available :http://www.jetir.org/papers/JETIR2601136.pdf

"XAI-Enabled Multi-Modal Text Analysis for Automated and Transparent Grading Systems", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 1, page no. ppb221-b224, January-2026, Available at : http://www.jetir.org/papers/JETIR2601136.pdf