SiDiaC-v.2.0: Sinhala Diachronic Corpus Version 2.0

SiDiaC-v.2.0 is the largest comprehensive Sinhala Diachronic Corpus to date, covering a period from 1800 CE to 1955 CE in terms of publication dates, and a historical span from the 5th to the 20th century CE in terms of written dates. The corpus consists of 244k words across 185 literary works that underwent thorough filtering, preprocessing, and copyright compliance checks, followed by extensive post-processing.

Texts from the National Library of Sri Lanka were selected from the SiDiaC-v.1.0 non-filtered list, which was digitised using Google Document AI OCR. This was followed by post-processing to correct formatting issues, address code-mixing, include special tokens, and fix malformed tokens. The construction of SiDiaC-v.2.0 was informed by practices from other corpora, such as FarPaHC, SiDiaC-v.1.0, and CCOHA.

Each work is categorised into two layers based on genre:

• Primary: binary — Fiction or Non-Fiction.
• Secondary: more detailed — Religious, History, Poetry, Language, Medical.

Paper: https://aclanthology.org/2026.lrec-1.537/ (DOI: 10.63317/2ybff4o55vrq)

GitHub Repository: https://github.com/NeviduJ/SiDiaC-v.2.0

Dataset composition

• Rows: 185
• Split: single train split
• Language: Sinhala (si)
• Total words: ~244k
• Publication date range: 1800 – 1955
• Written date range: 5th – 20th century CE
• Genres: Fiction, Non-Fiction, and their combinations with History, Poetry, Religious, Language, Medical
• OCR confidence range: 0.80 – 1.00

Schema

Field	Type	Description
title	string	Work title in Sinhala
title_en	string	Romanised/English transliteration of the title
author	string	Author name in Sinhala (or "Unknown")
author_en	string	Romanised/English transliteration of the author name
genre	string	Primary; Secondary genre categorisation (e.g. Fiction; Religious)
issued_date	string	Publication date
written_date	string	Composition period, where known
ocr_confidence	float	Heuristic OCR accuracy score for the finalised text
text	string	Finalised OCR-processed, post-processed text

Repository structure (source, on GitHub)

• Books_PDF/: Original source PDFs for each title.
• OCR_Final/: Finalised OCR results per title, each directory containing:
  • metadata.json: Bibliographic and processing metadata
  • <title>.txt: Plain text content extracted via OCR and post-processed

Example metadata.json

{
  "title": "පන්සිය පනස් ජාතක පොත",
  "title_en": "Pansiya Panas Jathaka Potha",
  "author": "Unknown",
  "author_en": "Unknown",
  "genre": "Fiction; Religious",
  "issued_date": "1881",
  "written_date": "1303 - 1333",
  "ocr_confidence": 0.9987
}

Usage

You can load this dataset using the datasets library:

from datasets import load_dataset

ds = load_dataset("Nevidu/SiDiaC-v.2.0", split="train")

# Verify the number of rows
print(f"Loaded dataset with {len(ds)} rows")

# Print the first example
print(ds[0])

Citation

@inproceedings{jayatilleke-etal-2026-sidiac,
    title = "{S}i{D}ia{C}-v.2.0: {S}inhala Diachronic Corpus Version 2.0",
    author = "Jayatilleke, Nevidu  and
      de Silva, Nisansa  and
      Sooriya-Arachchi, Uthpala Nimanthi  and
      Kulathilaka, Gagani Kasundhi  and
      Safrullah, Azra  and
      Sofalas, Johan Nevin",
    editor = "Piperidis, Stelios  and
      Bel, N{\'u}ria  and
      van den Heuvel, Henk  and
      Ide, Nancy  and
      Krek, Simon  and
      Toral, Antonio",
    booktitle = "Proceedings of the Fifteenth Language Resources and Evaluation Conference",
    month = may,
    year = "2026",
    address = "Palma de Mallorca, Spain",
    publisher = "ELRA Language Resource Association",
    url = "https://aclanthology.org/2026.lrec-1.537/",
    doi = "10.63317/2ybff4o55vrq",
    pages = "6740--6763",
    abstract = "SiDiaC-v.2.0 is the largest comprehensive Sinhala Diachronic Corpus to date, covering a period from 1800 CE to 1955 CE in terms of publication dates, and a historical span from the 5th to the 20th century CE in terms of written dates. The corpus consists of 244k words across 185 literary works that underwent thorough filtering, preprocessing, and copyright compliance checks, followed by extensive post-processing. Additionally, a subset of 59 documents totalling 70k words was annotated based on their written dates. Texts from the National Library of Sri Lanka were selected from the SiDiaC-v.1.0 non-filtered list, which was digitised using Google Document AI OCR. This was followed by post-processing to correct formatting issues, address code-mixing, include special tokens, and fix malformed tokens. The construction of SiDiaC-v.2.0 was informed by practices from other corpora, such as FarPaHC, SiDiaC-v.1.0, and CCOHA. This was particularly relevant for syntactic annotation and text normalisation strategies, given the shared characteristics of low-resource language status between Faroese and the similar cleaning strategies utilised in CCOHA. This corpus is categorised into two layers based on genres: primary and secondary. The primary categorisation is binary, assigning each book to either Non-Fiction or Fiction. The secondary categorisation is more detailed, grouping texts under specific genres such as Religious, History, Poetry, Language, and Medical. Despite facing challenges due to limited resources, SiDiaC-v.2.0 serves as a comprehensive resource for Sinhala NLP, building upon the work previously done in SiDiaC-v.1.0.",
}