Kratos-AI/Bengali_Documents_Dataset_PDF

Bengali Documents Dataset (PDF)

This dataset contains a curated collection of Bengali-language documents in PDF format. The corpus includes educational materials, books, reports, articles, and public-domain texts written in Bengali script. It is designed to support AI research in OCR, document understanding, and text extraction for Indic languages.

Contact

For queries or collaborations related to this dataset, contact:

• [email protected]
• [email protected]

Supported Tasks

• Task Categories:

  • Document Classification
  • OCR and Text Recognition
  • Layout and Structure Analysis
  • Language Modeling for Bengali

• Supported Tasks:

  • Extraction of Bengali text from PDF documents
  • Classification by topic or domain (education, literature, news, government)
  • OCR for printed Bengali script and mixed-language layouts
  • Benchmarking Bengali-language document processing models

Languages

• Primary Language: Bengali
• Secondary Presence: English terms, numerals, and section headers (common in bilingual publications)

Dataset Creation

Curation Rationale

The dataset was curated to advance AI systems that process Bengali-language documents, enhance OCR quality for Indic scripts, and enable structured understanding of mixed-script PDFs.

Source Data

• Contributors: Public archives, open-access Bengali libraries, educational repositories, and volunteer submissions.
• Collection Process: Documents were compiled from freely available PDF sources with permissible open licenses.

Other Known Limitations

• Bias: Focused more on educational and literary content; limited representation of informal or handwritten material
• Format Variation: Some PDFs include embedded images or scanned pages
• Dialect Coverage: Primarily Standard Bengali (বাংলা), limited regional dialects

Intended Uses

✅ Direct Use

• Training OCR and NLP systems for Bengali document understanding
• Research on multilingual document parsing and Indic language digitization
• Educational and academic AI research for Bengali text extraction

❌ Out-of-Scope Use

• Extracting or profiling personal data from documents
• Commercial reuse of copyrighted content
• Use in surveillance, handwriting identification, or behavioral analysis

License

CC BY 4.0