Update README.md

README.md

@@ -1,3 +1,83 @@
----
-license: cc-by-4.0
----
+---
+license: cc-by-4.0
+pretty_name: 'Korean Documents Dataset'
+language:
+- ko
+tags:
+- document
+- korean
+- text-recognition
+- ocr
+- scanned-text
+- handwriting
+- printed-text
+- ai-research
+- document-understanding
+- nlp
+- computer-vision
+task_categories:
+- document-classification
+- text-recognition
+- document-understanding
+size_categories:
+- 1K<n<10K
+---
+
+# Korean Documents Dataset
+*This dataset contains a curated collection of scanned and digital Korean documents, including printed materials, forms, letters, notices, and handwritten notes. It has been cleaned and anonymized for safe use in AI training, OCR benchmarking, and natural language understanding research.*
+
+## Contact
+For queries or collaborations related to this dataset, contact:  
+  - anoushka@kgen.io  
+  - abhishek.vadapalli@kgen.io  
+
+## Supported Tasks
+
+- **Task Categories**:  
+  - Document Classification  
+  - Text Recognition (OCR)  
+  - Document Understanding  
+
+- **Supported Tasks**:  
+  - Optical character recognition for printed and handwritten Korean text  
+  - Document layout and structure analysis  
+  - Form parsing and entity extraction  
+  - Language modeling for Korean official or administrative text  
+  - Multimodal document understanding (text + layout)  
+  - Translation benchmarking (Korean ↔ English)  
+
+## Languages
+
+- **Primary Language**: Korean  
+- **Secondary Presence**: Limited English in bilingual official documents  
+
+## Dataset Creation
+
+### Curation Rationale
+The dataset was created to improve AI systems’ ability to read, understand, and extract structured information from Korean documents. It serves as a benchmark for OCR, NLP, and layout recognition tasks for Korean-language content.
+
+### Source Data
+- **Contributors**: Digitized archives, simulated official forms, and user-submitted handwritten samples  
+- **Collection Process**: Each document was scanned or photographed, then manually reviewed to ensure removal of any personally identifiable information (PII). The dataset includes both structured (forms, applications) and unstructured (letters, notes) materials.  
+
+### Other Known Limitations
+- **Bias**: Majority of documents are formal or administrative; limited representation of casual or creative writing  
+- **Quality Variability**: Some handwritten samples may include unclear or stylized writing  
+- **Format Range**: Primarily PDFs or images; minimal availability of native digital text files  
+
+## Intended Uses
+
+### ✅ Direct Use
+- Training OCR and document understanding models for Korean  
+- Benchmarking handwriting recognition systems  
+- Academic research in NLP and document layout analysis  
+- Translation model training (Korean ↔ English)  
+
+### ❌ Out-of-Scope Use
+- Reconstructing real personal or institutional data  
+- Commercial reuse of document content without proper licensing  
+- Any use involving personal or confidential information reconstruction  
+
+## License
+
+CC BY 4.0