README.md

metadata

pretty_name: Indic Contextual Post-OCR Correction (Hindi, Gujarati, Marathi)
language:
  - hi
  - gu
  - mr
multilinguality: multilingual
task_categories:
  - text-generation
tags:
  - ocr
  - post-ocr-correction
  - error-correction
  - text-correction
  - indic
  - devanagari
  - hindi
  - gujarati
  - marathi
license: cc-by-4.0
size_categories:
  - 100K<n<1M
configs:
  - config_name: hi
    data_files:
      - split: train
        path: train-hi.csv
      - split: validation
        path: val-hi.csv
      - split: test
        path: test-hi.csv
  - config_name: mr
    data_files:
      - split: train
        path: train-mr.csv
      - split: validation
        path: val-mr.csv
      - split: test
        path: test-mr.csv
  - config_name: gu
    data_files:
      - split: train
        path: train-gu.csv
      - split: validation
        path: val-gu.csv
      - split: test
        path: test-gu.csv

Indic Contextual Post-OCR Correction

Dataset Summary

This dataset supports contextual post-OCR correction for Indic languages. Each example is a sentence-level triple consisting of:

• an OCR-generated sentence (noisy),
• the preceding sentence used as context, and
• the corrected sentence (ground truth).

Hugging Face dataset page: https://huggingface.co/datasets/AbhishekBhandari/Indic-post-ocr-correction

Supported Tasks

• Post-OCR text correction (sentence-level)
• Context-aware text generation for correcting OCR noise
• Evaluation with/without preceding-sentence context

Languages

• Hindi (hi) — Devanagari
• Gujarati (gu)
• Marathi (mr) — Devanagari

Dataset Structure

Configs (language-wise subsets)

This dataset is organized into three configs:

• hi (Hindi)
• mr (Marathi)
• gu (Gujarati)

Each config contains train, validation, and test splits.

Data Instances

A typical instance:

{
  "ocr_sentence": "...",
  "context_sentence": "...",
  "ground_truth": "..."
}

Citation

If you use this dataset, please cite the following paper:

@article{bhandari2026framework,
  title={A Framework and Dataset for Contextual Post-OCR Correction},
  author={Bhandari, Abhishek and Harit, Gaurav},
  journal={ACM Transactions on Asian and Low-Resource Language Information Processing},
  volume={25},
  number={6},
  pages={1--20},
  year={2026},
  publisher={ACM New York, NY}
}