sebiena/amharic_emotion_detection

Amharic social media text dataset labeled with six Ekman emotions (anger, disgust, fear, joy, sadness, surprise), adapted to domestic context. Prepared for training and evaluation of fine-tuned emotion detection models. # amharic_emotion_detection Amharic social media text dataset labeled with six Ekman emotions (anger, disgust, fear, joy, sadness, surprise), adapted to domestic context. Prepared for training and evaluation of **fine-tuned** emotion detection models. --- ## 📂 Dataset Overview - **Language**: Amharic - **Source**: Social media content - **Labels**: Six Ekman emotions - Anger - Disgust - Fear - Joy - Sadness - Surprise - **Use Case**: Emotion tone prediction, NLP research, fine-tuning transformer-based models --- ## 🔧 Usage You can use this dataset for: - Training baseline models (e.g., SVM, RF, CNN, Bi-LSTM) - Fine-tuning transformer models (e.g., BERT, RoBERTa, XLM-R) - Evaluating emotion detection systems in Amharic Example (Python + Pandas): ```python import pandas as pd # Load dataset df = pd.read_csv("amharic_emotion_dataset.csv") # Preview print(df.head())