PatrickIIT/NLP-NER

This project focuses on fine-tuning a pre-trained RoBERTa model (benjamin/roberta-base-wechsel-swahili) for Named Entity Recognition (NER) in Swahili using the MasakhaNER dataset. The goal is to identify and classify named entities such as PERSON, LOCATION, ORGANIZATION, DATE, etc., in Swahili text. # **Named Entity Recognition (NER) for Swahili using RoBERTa-Base-Wechsel** ## 📌 Table of Contents - Overview - Dataset - Environment Setup - Preprocessing - Model Training - Validation with Swahili Speakers - Evaluation Metrics - Results and Discussion - How to Use - Contributing - License - Acknowledgments --- ## 🚀 Overview This project fine-tunes **RoBERTa-base-Wechsel-Swahili** for **Named Entity Recognition (NER) in Swahili**, using the **MasakhaNER dataset**. The goal is to identify and classify named entities such as **Persons (PER), Locations (LOC), Organizations (ORG), and Dates (DATE)** in Swahili text. ### **Key Features:** ✅ Fine-tuning **RoBERTa-base-Wechsel-Swahili** for Swahili NER. ✅ **Linguistic validation** with Swahili speakers for accuracy. ✅ **Evaluation using Precision, Recall, F1-score, and Accuracy.** ✅ **Error analysis and optimization suggestions.** --- ## 📊 Dataset The dataset used in this project is **MasakhaNER**, which contains Swahili text annotated with named entities. ### **Dataset Statistics:** | **Subset** | **Number of Sentences** | |-------------|-----------------------| | **Train** | 2,109 | | **Validation** | 300 | | **Test** | 604 | Each sentence consists of: - `id`: Unique identifier - `tokens`: Words/tokens in the sentence - `ner_tags`: Named entity labels (`O`, `B-PER`, `I-PER`, `B-LOC`, etc.) ### **Load the Dataset** ```python from datasets import load_dataset dataset = load_dataset("masakhaner", "swa", trust_remote_code=True) Each example consists of: - `id`: Unique identifier for the example. - `tokens`: List of words/tokens in the sentence. - `ner_tags`: Corresponding labels for each token (e.g., `O`, `B-PER`, `I-PER`, `B-LOC`, etc.). ``` --- ## Environment Setup To run this project, ensure you have the following dependencies installed: ### Required Libraries ```bash pip install transformers datasets seqeval evaluate ``` ### Python Version This pr …