AnyibabaIG/Igbo-Proverbs-NLP

Tone-marked linguistic data for low-resource NLP. # 🇳🇬 Tone-Marked Igbo Proverbs Parallel Corpus This repository hosts the documentation, research context and linguistic frameworks for the **Igbo Proverbs Dataset**. This project serves as a highly specialised, low-resource language resource designed to bridge the gap between traditional African oral literature and modern Natural Language Processing (NLP). ## Project Overview Rather than focusing on sheer volume, this dataset acts as a deeply curated **linguistic micro-corpus**. It provides a granular, multidimensional look at structural variations, tonal inflections, and contextual applications within localised Igbo dialects. It establishes a high quality data baseline meant for fine-tuning linguistic models, testing AI translation systems, and preserving indigenous knowledge systems. ## Dataset Access To protect the integrity of this specialised data and track its academic or commercial application, the primary dataset file is securely hosted and gated on Hugging Face. 👉 **Download the Dataset on Hugging Face** *Note: Access is granted automatically. You simply need to log in with a Hugging Face account and quickly fill out the short form stating your intended project or research use case.* ## 📋 Dataset Specifications - **Core Language:** Igbo (`ig`) - **Dataset Volume:** 50 highly curated entries (Ongoing compilation) - **Format Type:** Structured CSV - **License Framework:** CC-BY-4.0 (Attribution Required) ### Rich 10-Unit Data Schema To ensure maximum analytical depth, every single proverb row is mapped across 10 precise linguistic and contextual dimensions: 1. **Proverb Code:** Unique identifier for systematic indexing and cross-referencing. 2. **Raw Dialectal Proverb:** The original proverb preserved with native, localised pronunciation and tone markings. 3. **Standard Orthography:** The proverb structurally mapped to central standard Igbo orthography with full tone markings. 4. **Literal Meaning:** Word-for-word translation to assist comparative sem …