lojban / README.md
damfle's picture
doc: README.md
ae43163 verified
metadata
license: pddl
language:
  - jbo
task_categories:
  - text-generation
tags:
  - lojban
  - jbo
pretty_name: lojban
size_categories:
  - 1K<n<10K

Lojban Dataset

This is a curated collection of Lojban language resources merged from multiple sources including finewiki, fineweb2, finepdf, and official Lojban resources.

Overview

Lojban is a constructed language designed to be more logically precise and unambiguous than natural languages. This dataset brings together various Lojban texts and resources in a unified format for machine learning, linguistic research, and language model training.

Data Sources

This dataset is a merge of:

  • finewiki - Wikipedia-derived content adapted for Lojban
  • fineweb2 - Web-sourced Lojban texts and documentation
  • finepdf - PDF documents converted to text format
  • Official Lojban Resources - Materials from the official Lojban website and community resources

Dataset Format

The dataset is provided in Parquet format (lojban.parquet) for efficient storage and processing.

Usage

from datasets import load_dataset

dataset = load_dataset("huggingface", "lojban")

Features

  • Language: Lojban (la .lojban.)
  • Format: Parquet (columnar storage format)
  • Size: See dataset card for specific statistics
  • Quality: Curated from multiple authoritative sources

About Lojban

Lojban is a logical language created in the 1980s, designed to test the linguistic relativity hypothesis by removing grammatical ambiguities found in natural languages. Key characteristics:

  • Logical Structure: Grammar is fully specified with no ambiguities
  • Unambiguous: Each utterance has exactly one parse tree
  • Constructed: Artificially created (not a natural language)
  • Active Community: Maintained by dedicated speakers worldwide

Use Cases

  • Linguistic research and computational linguistics
  • Training language models for low-resource languages
  • Studying constructed languages
  • Logical language processing
  • Natural language ambiguity research

License

Please refer to the original source licenses for finewiki, fineweb2, and finepdf components, as well as the official Lojban resources license terms.

Resources

Contributing

For issues, suggestions, or contributions related to this dataset, please open an issue on the Hugging Face dataset hub.


Version: 1.0 Last Updated: February 2026