Skip to main content Skip to journal navigation menu Skip to site footer
contact@ukscip.com
Article

A Divergence-Driven Framework for Chinese Literary Text Digitization: Multi-Engine OCR Accuracy Evaluation and Multi-LLM Post-Correction

Xin Zhang ORCID ,  Rachel Suet Kay Chan ORCID ,  Raan-Hann Tan ORCID
    Received: 4 May 2026; Revised: 26 June 2026; Accepted: 6 July 2026; Published: 21 September 2026

    Abstract

    The digitization of Chinese literary texts for corpus construction faces three challenges: vertical typesetting, a set covering more than 80,000 Chinese characters codified in China’s national standard GB 18030, and frequent use of rare characters. To address gaps in existing research on multi-engine comparison and accuracy quantification, this study adopts a three-phase framework: engine evaluation, workflow design, and end-to-end validation. Engine evaluation comprises two experiments. Experiment 1, on 13 pages of vertical Traditional Chinese, compares three engines: Kandian Guji v2.0.7 (Kandian) achieves 0.25% Character Error Rate (CER), far surpassing PaddleOCR-VL-1.5 (Paddle1.5) at 4.77% and ABBYY FineReader 15 (ABBYY) at 5.96%. Experiment 2 extends to four engines on 21 pages of horizontal Simplified Chinese by adding PP-OCRv5 (PP5); Paddle1.5 achieves the lowest CER at 0.74%, and its error distribution diverges across all three error classes from Kandian’s, supporting a multi-engine scheme. ABBYY and PP5 are excluded from the final workflow due to error patterns. The four-stage workflow is adopted for the Simplified Chinese scenario: multi-engine Optical Character Recognition (OCR), character-level diff detection, multi-Large Language Model (LLM) arbitration, and final human proofreading. Experiment 3 validates the end-to-end workflow on 20 pages of Simplified Chinese text, where Kandian and Paddle1.5 produce initial CERs of 0.54% and 0.28%; diff detection identifies 62 divergences, of which 49 reach consensus, reducing CER to 0.26%; human proofreading reduces it to 0.00%. For the Traditional Chinese scenario, a rule-based script grounded in character co-occurrence legitimacy is adopted, reducing CER from 0.286% to 0.074%. The framework offers a reproducible quality-control method for Chinese literary digitization.

    Keywords

    References

    1. 1. Eren, F.; Çay, K. Elements of a Digital Urbanisation Strategy for Türkiye: Evidence from Psychometric Testing. Soc. Sci. 2025, 14, 89.
    2. 2. Li, C.; Guo, R.; Zhou, J.; et al. PP-StructureV2: A Stronger Document Analysis System. arXiv preprint 2022, arXiv:2210.05391. DOI: https://doi.org/10.48550/arXiv.2210.05391
    3. 3. Standardization Administration of the People’s Republic of China. GB 18030-2022: Information Technology—Chinese Coded Character Set; China Standard Press: Beijing, China, 2022.
    4. 4. The Unicode Consortium. The Unicode Standard, Version 17.0.0; Unicode, Inc.: Mountain View, CA, USA, 2025.
    5. 5. Holley, R. How Good Can It Get?: Analysing and Improving OCR Accuracy in Large Scale Historic Newspaper Digitisation Programs. D-Lib Mag. 2009, 15. DOI: https://doi.org/10.1045/march2009-holley
    6. 6. Neudecker, C.; Baierer, K.; Gerber, M.; et al. A Survey of OCR Evaluation Tools and Metrics. In Proceedings of the 6th International Workshop on Historical Document Imaging and Processing, Lausanne, Switzerland, 5–6 September 2021; pp. 13–18. DOI: https://doi.org/10.1145/3476887.3476888
    7. 7. Greif, G.; Griesshaber, N.; Greif, R. Multimodal LLMs for OCR, OCR Post-Correction, and Named Entity Recognition in Historical Documents. arXiv preprint 2025, arXiv:2504.00414. DOI: https://doi.org/10.48550/arXiv.2504.00414
    8. 8. Wang, K.; Yang, F.; Jiang, S. Review of Optical Character Recognition. Appl. Res. Comput. 2020, 37, 22–24. (in Chinese)
    9. 9. Chen, X.; Jin, L.; Zhu, Y.; et al. Text Recognition in the Wild: A Survey. ACM Comput. Surv. 2021, 54, 42. DOI: https://doi.org/10.1145/3440756
    10. 10. Li, M.; Lv, T.; Chen, J.; et al. TrOCR: Transformer-Based Optical Character Recognition with Pre-Trained Models. Proc. AAAI Conf. Artif. Intell. 2023, 37, 13094–13102. DOI: https://doi.org/10.1609/aaai.v37i11.26538
    11. 11. Cao, N.A. Resources and Tools for Corpus Compilation of Translated Literary Texts in Late Qing and Republican Period. J. Transl. Stud. 2018, 2, 153–168.
    12. 12. Liu, C.; Jian, C.; Huang, J.; et al. A Robust and Efficient Algorithm for Chinese Historical Document Analysis and Recognition. Natl. Sci. Rev. 2023, 10, nwad115. DOI: https://doi.org/10.1093/nsr/nwad115
    13. 13. Wong, N.Y.H. Modernism’s Vector: Dataset for a Chinese-Language Journal in Late 1950s Malaya. J. Open Humanit. Data 2026, 12, 40. DOI: https://doi.org/10.5334/johd.485
    14. 14. Zavorin, I.; Borovikov, E.; Borovikov, A.; et al. A Multi-Evidence, Multi-Engine OCR System. In Proceedings of the Document Recognition and Retrieval XIV, San Jose, CA, USA, 30 January–1 February 2007. DOI: https://doi.org/10.1117/12.703106
    15. 15. Nguyen, T.T.H.; Jatowt, A.; Coustaty, M.; et al. Survey of Post-OCR Processing Approaches. ACM Comput. Surv. 2022, 54, 1–37. DOI: https://doi.org/10.1145/3453476
    16. 16. Ramirez-Orta, J.A.; Xamena, E.; Maguitman, A.; et al. Post-OCR Document Correction with Large Ensembles of Character Sequence-to-Sequence Models. Proc. AAAI Conf. Artif. Intell. 2022, 36, 11192–11199. DOI: https://doi.org/10.1609/aaai.v36i10.21369
    17. 17. Rijhwani, S.; Rosenblum, D.; Anastasopoulos, A.; et al. Lexically Aware Semi-Supervised Learning for OCR Post-Correction. Trans. Assoc. Comput. Linguist. 2021, 9, 1285–1302. DOI: https://doi.org/10.1162/tacl_a_00427
    18. 18. Thomas, A.; Gaizauskas, R.; Lu, H. Leveraging LLMs for Post-OCR Correction of Historical Newspapers. In Proceedings of the Third Workshop on Language Technologies for Historical and Ancient Languages, Torino, Italy, May 2024; pp. 116–121.
    19. 19. Kanerva, J.; Ledins, C.; Käpyaho, S.; et al. OCR Error Post-Correction with LLMs in Historical Documents: No Free Lunches. arXiv preprint 2025, arXiv:2502.01205. DOI: https://doi.org/10.48550/arXiv.2502.01205
    20. 20. Kettunen, K.; Koistinen, M.; Kervinen, J. Ground Truth OCR Sample Data of Finnish Historical Newspapers and Journals in Data Improvement Validation of a Re-OCRing Process. LIBER Q. 2020, 30, 1–20. DOI: https://doi.org/10.18352/lq.10322
    21. 21. Cui, C.; Sun, T.; Liang, S.; et al. PaddleOCR-VL-1.5: Towards a Multi-Task 0.9B VLM for Robust In-the-Wild Document Parsing. arXiv preprint 2026, arXiv:2601.21957. DOI: https://doi.org/10.48550/arXiv.2601.21957
    22. 22. Li, Y. Sea-East Green: A Parable of Taipei; Unitas Publishing: Taipei, Taiwan, 1992. (in Chinese)
    23. 23. Redfield, J. The Tenth Insight: Holding the Vision; Li, Y., Trans.; Yuan-Liou Publishing: Taipei, Taiwan, 1997. (in Chinese)
    24. 24. Kübler-Ross, E. The Wheel of Life: A Memoir of Living and Dying; Li, Y., Trans.; Commonwealth Publishing: Taipei, Taiwan, 1998. (in Chinese)
    25. 25. Mo, Y. Red Sorghum: A Novel of China; PLA Literature and Arts Publishing House: Beijing, China, 1987. (in Chinese)
    ×