AI Exam Controller for Question Paper Generation and Answer Sheet Evaluation with Secure Result Processing

References

[1] Choi, J., Hong, S., Park, J., & Jung, E. S. (2025). Toward generating quality test questions and answers using quantized low-rank adapters in large language models. IEEE Access, 13, 87793–87810.

[2] Tripathi, S., Nafis, M. T., Hussain, I., & Saudagar, A. K. J. (2025). Multimodal fine-tuning of large language models for robust document visual question answering. IEEE Access, 13, 174611–174623.

[3] Shoaib, M., Husnain, G., Sayed, N., & Qahmash, A. (2025). Automated generation of multiple-choice questions for computer science education using conditional generative adversarial networks. IEEE Access, 13, 16697–16715.

[4] Duong, T. N. B., & Shar, L. K. (2025). GenScore: Agent-based short-answer question generation and scoring in software engineering courses. IEEE Access, 13, 215804–215820.

[5] Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. Proceedings of NAACL-HLT, Minneapolis, USA, 4171–4186.

[6] Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., & Liu, P. J. (2020). Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research, 21(1), 1–67.

[7] Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., & Polosukhin, I. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30, 5998–6008.

[8] Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., & Amodei, D. (2020). Language models are few-shot learners. Advances in Neural Information Processing Systems, 33, 1877–1901.

[9] Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., & Chen, W. (2022). LoRA: Low-rank adaptation of large language models. Proceedings of ICLR, Virtual Conference, 1–13.

[10] Yang, Z., Xu, Y., Cui, L., Zhang, S., Wei, F., Wang, W., & Liu, T. (2020). LayoutLM: Pre-training of text and layout for document image understanding. Proceedings of KDD, Virtual Conference, 1192–1200.

[11] Xu, Y., Xu, Y., Lv, T., Cui, L., Wei, F., Wang, G., & Liu, T. (2021). LayoutLMv2: Multi-modal pre-training for visually rich document understanding. Proceedings of ACL, Online, 2579–2591.

[12] Manning, C. D., Raghavan, P., & Schütze, H. (2008). Introduction to Information Retrieval. Cambridge University Press, Cambridge, UK.

[13] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press, Cambridge, Massachusetts, USA.

[14] Jurafsky, D., & Martin, J. H. (2024). Speech and Language Processing: An Introduction to Natural Language Processing (4th ed.). Pearson Education, New Jersey, USA.

[15] Nakamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system. Cryptography Mailing List, 1–9. Retrieved from https://bitcoin.org/bitcoin.pdf

@article{1715500,
      author = {M Purushothaman, D. Revathy},
      title = {AI Exam Controller for Question Paper Generation and Answer Sheet Evaluation with Secure Result Processing},
      journal = {Iconic Research And Engineering Journals},
      year = {2026},
      volume = {9},
      number = {9},
      pages = {2471-2477},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1715500.pdf},
      abstract = {Contemporary university examination systems face persistent challenges including inconsistent evaluation, evaluator fatigue, marks tampering, and delayed result publication. This paper presents an AI-driven Examination Controller System that automates the complete examination lifecycle from question generation to secure result publication. Subject PDFs are processed using PyMuPDF, with key concepts extracted through TF-IDF and TextRank algorithms. The T5 Transformer synthesizes extracted concepts into syllabus-aligned examination questions structured into standardized formats. Post-examination, scanned answer sheets undergo OpenCV preprocessing and Tesseract OCR-based text extraction. Student responses are encoded using BERT embeddings and evaluated against reference content through cosine similarity, enabling context-aware, bias-free mark allocation. Evaluated marks are cryptographically secured within a SHA-256 blockchain ledger, rendering tampering immediately detectable. A Controller of Examinations verification workflow ensures institutional oversight before result publication. The system significantly reduces manual workload, improves evaluation consistency, and demonstrates strong scalability for large examination ecosystems including autonomous universities and affiliated institutions.},
      keywords = {T5 Transformer, BERT Embeddings, Cosine Similarity, OCR, Blockchain, Examination Management, NLP},
      month = {March},
      doi = {https://doi.org/10.64388/IREV9I9-1715500}
  }