Update README.md

README.md

@@ -1,122 +1,144 @@
 ---
 license: cc-by-4.0
 ---
+
 # Eng-PidginEdu Dataset
 
 ## Overview
 
-Eng-PidginEdu is a large-scale English–Nigerian Pidgin parallel corpus designed to support machine translation and educational accessibility research. The dataset contains **26,232 sentence pairs** spanning **8 core secondary school subjects**, with approximately **1.28 million tokens** and **99.997% alignment quality**.
+Eng-PidginEdu is an English–Nigerian Pidgin parallel corpus developed to support machine translation, multilingual NLP, and educational accessibility research for low-resource African languages.
+
+The dataset contains **26,232 parallel sentence pairs** collected across **8 Nigerian secondary school subjects**, comprising approximately **1.09 million total tokens** with high-quality sentence-level alignment and human-validated translations.
 
-This dataset addresses the critical gap in educational resources for Nigerian Pidgin, a lingua franca spoken by over 75 million people, enabling the development of AI systems that improve access to education across linguistically diverse regions.
+Nigerian Pidgin serves as a major lingua franca spoken across Nigeria and West Africa. This dataset was created to improve access to educational technologies and language resources for Pidgin-speaking communities.
 
 ---
 
-## Dataset Composition
+# Dataset Composition
 
-* **Total sentence pairs:** 30,618
+## Statistics
 
-* **Subjects covered:**
+| Metric | Value |
+|---|---|
+| Total Sentence Pairs | 26,232 |
+| English Tokens | 555,398 |
+| Pidgin Tokens | 533,657 |
+| Avg English Sentence Length | 21.17 tokens |
+| Avg Pidgin Sentence Length | 20.34 tokens |
+| English TTR | 0.1032 |
+| Pidgin TTR | 0.0883 |
 
-  * Computer Science
-  * Business Studies
-  * Government
-  * Civic Education
-  * Social Studies
-  * Biology
-  * English Language
-  * History
 
-* **Splits:**
+---
 
-  * Train: 24,493
-  * Validation: 3,061
-  * Test: 3,063
+## Dataset Splits
 
-* **Languages:**
+| Split | Size |
+|---|---:|
+| Train | 20,985 |
+| Validation (Dev) | 2,623 |
+| Test | 2,624 |
 
-  * Source: English (`en`)
-  * Target: Nigerian Pidgin (`pcm`)
+The dataset was randomly shuffled using a fixed seed (`42`) to ensure reproducibility.
 
 ---
 
-## Data Collection Process
-
-The dataset was constructed from publicly accessible Nigerian secondary school textbooks and open educational resources aligned with national curriculum standards. Texts were digitized and processed through:
+# Languages
 
-1. Automated text extraction (PyPDF2, python-docx)
-2. Sentence segmentation (NLTK punkt tokenizer)
-3. Multi-stage cleaning (spaCy, textacy, regex filtering)
+| Role | Language | Code |
+|---|---|---|
+| Source | English | `en` |
+| Target | Nigerian Pidgin | `pcm` |
 
-Translations were generated using a hybrid pipeline:
+---
 
-* Machine translation using **Davlan/mt5-small-en-pcm**
-* Rule-based quality filtering
-* Human-in-the-loop validation by native Nigerian Pidgin speakers
-* Glossary based augumentation
+# Data Collection Process
 
-This process ensured high-quality, educationally faithful translations.
+The dataset was compiled from publicly accessible Nigerian secondary school educational materials and curriculum-aligned learning resources.
 
----
+The data preparation pipeline included:
 
-## Key Features
+1. Document extraction using:
+   - `PyPDF2`
+   - `python-docx`
 
-* Large-scale educational parallel corpus for a low-resource language
-* Multi-domain coverage across 8 subjects
-* Human-validated translations
-* Support for code-switching and natural Pidgin variation
-* Designed for real-world educational deployment
+2. Sentence segmentation using:
+   - NLTK Punkt tokenizer
 
----
+3. Text normalization and cleaning using:
+   - spaCy
+   - textacy
+   - regular-expression filtering
 
-## Intended Use
+4. Translation generation using:
+   - `Davlan/mt5-small-en-pcm`
 
-This dataset is intended for:
+5. Human validation and correction by native Nigerian Pidgin speakers
 
-* Machine translation (English ↔ Nigerian Pidgin)
-* Educational AI systems
-* Low-resource NLP research
-* Multilingual learning platforms
+6. Alignment verification and post-processing
 
 ---
 
-## Limitations
+# Key Features
 
-* Domain-specific (education-focused)
-* Contains orthographic variation in Pidgin
-* Includes code-switching, which may affect standard evaluation metrics
-* Not representative of all dialectal variations of Nigerian Pidgin
+- Large-scale English–Pidgin educational parallel corpus
+- Coverage across multiple academic subjects
+- Human-validated translations
+- Natural Nigerian Pidgin expressions and code-switching
+- Suitable for low-resource machine translation research
+- Reproducible train/dev/test splits
 
 ---
 
-## Bias and Ethical Considerations
+# Intended Use
 
-* Biased toward formal educational language
-* Uneven subject distribution (e.g., Computer Science overrepresented)
-* Reflects real-world linguistic variation rather than standardized forms
+This dataset is intended for:
 
-No personal or sensitive data is included.
+- Machine Translation (MT)
+- Low-resource NLP research
+- Educational AI systems
+- Cross-lingual learning applications
+- Multilingual language modeling
+- Evaluation of African language translation systems
 
 ---
 
-## Licensing
+# Limitations
 
-This dataset is released under the **CC-BY-4.0 License**.
+- Educational domain only
+- Uneven subject distribution
+- Contains orthographic variation in Nigerian Pidgin
+- Includes informal and code-switched language patterns
+- Does not represent all regional varieties of Nigerian Pidgin
 
 ---
 
-## Citation
+# Bias and Ethical Considerations
 
-If you use this dataset, please cite:
+The dataset reflects naturally occurring educational language and may contain:
 
-Oladipupo, F. (2026). *Eng-PidginEdu: Scaling Nigerian Pidgin Educational Content through Community-Driven Machine Translation*. 
+- Curriculum-specific biases
+- Subject imbalance
+- Linguistic variation associated with Nigerian Pidgin usage
+
+No personally identifiable or sensitive information is included.
 
 ---
 
-## Acknowledgements
+# Licensing
 
-We acknowledge the contributions of native Nigerian Pidgin speakers who participated in validation, and the open educational resources that made this dataset possible.
+This dataset is released under the **CC-BY-4.0 License**.
 
 ---
 
+# Citation
 
+If you use this dataset, please cite:
 
+```bibtex
+@dataset{eng_pidginedu_2026,
+  author = {Oladipupo, F.},
+  title = {Eng-PidginEdu: An English–Nigerian Pidgin Educational Parallel Corpus},
+  year = {2026},
+  publisher = {Hugging Face}
+}