AI-Based Question Paper Generation and Quality Optimization: An Integrated Computational Approach to Intelligent Assessment Construction

DOI

https://www.doi.org/10.59256/indjcst.20260502085

Published In

Volume 5, Issue 2 (2026)

Published Online

May-August 2026

Pages

753-760

Article Type

Original Article

Check for updates

Academic examinations play a fundamental role in evaluating student learning outcomes and maintaining educational standards. However, the quality of examination papers often depends on manual preparation by subject experts, resulting in inconsistencies in cognitive complexity, syllabus coverage, difficulty distribution, linguistic quality, and fairness. To address these challenges, this paper presents AIQPGQO (AI-Based Question Paper Generation and Quality Optimization), an integrated framework for intelligent question paper generation and comprehensive quality evaluation. The proposed framework combines knowledge graph-based concept modelling, large language model- driven question generation, Bloom's Revised Taxonomy alignment, psychometric difficulty calibration, reinforcement learning-based paper assembly, automated fairness assessment, and continuous feedback adaptation within a unified architecture. Unlike conventional automated question generation systems that primarily focus on producing individual questions, AIQPGQO optimizes the complete examination paper by simultaneously considering cognitive balance, syllabus coverage, diversity, linguistic correctness, and institutional assessment requirements. The framework was evaluated using a multi-institutional dataset consisting of 1,490 examination papers containing approximately 30,800 annotated questions collected from six academic disciplines. Experimental evaluation demonstrates that the proposed approach consistently outperforms existing baseline methods in cognitive level classification, difficulty estimation, syllabus coverage analysis, linguistic quality assessment, and overall paper quality verification. Ablation studies further confirm the contribution of each architectural component, while cross-domain experiments demonstrate strong generalization across diverse academic disciplines. The results indicate that AIQPGQO provides a practical and scalable solution for intelligent examination design, supporting educational institutions in developing reliable, balanced, and high-quality assessment papers while reducing manual effort and improving the consistency, transparency, and fairness of the examination process.