CharlesMoslonka/finance_pdfs_OCR_tests

OCR'd Financial Documents Dataset

A comprehensive dataset of OCR'd financial documents (annual reports, SEC filings) from various stock exchanges including NYSE, NASDAQ, LSE, ASX, TSX, and OTC markets.

Dataset Structure

dataset-root/
├── documents/
│   ├── NYSE_DEI_2019/
│   │   ├── document.pdf                 # Original PDF file
│   │   ├── metadata.json                # Document metadata
│   │   ├── document.html                # OCR output (HTML format)
│   │   ├── document.md                  # OCR output (Markdown format)
│   │   └── images/
│   │       ├── <hash>_1_img.webp        # Page 1 image
│   │       ├── <hash>_2_img.webp        # Page 2 image
│   │       └── ...
│   ├── NASDAQ_CASS_2022/
│   │   └── ...
│   └── ... (100 documents total)
├── dataset_info.json                    # Dataset metadata
└── README.md

Document Organization

Each document folder contains:

• PDF file (<document_name>.pdf): The original PDF document
• Metadata (<document_name>_metadata.json): Document metadata including source, date, and OCR information
• OCR Output (.html and .md files): Text extracted via OCR
• Images (images/ folder): WebP images of each page, named as <hash>_<page_number>_img.webp

Document Information

• Total Documents: 100
• Stock Exchanges: NYSE, NASDAQ, LSE, ASX, TSX, OTC
• Time Period: 2000-2024
• Image Format: WebP
• OCR Type: Document OCR with image extraction

File Naming

Document folders follow the pattern: {EXCHANGE}_{TICKER}_{YEAR}

• Example: NYSE_DEI_2019 (NYSE, ticker DEI, year 2019)
• Example: NASDAQ_CASS_2022 (NASDAQ, ticker CASS, year 2022)