Spam Detection & Emotion Analysis with DL models

Struggling with spam or deciphering online sentiment? This repository delves into text classification using two powerful deep learning techniques: LSTM and fine-tuned DistilBERT. Explore how these models tackle real-world tasks like detecting unwanted emails and identifying emotions in tweets.

Installation

1. Clone the repository:

git clone https://github.com/hoverslam/nlp-text-classification/

2. Install the dependencies:

pip install -r requirements.txt

3. Run the notebooks:
   • spam_detection
   • emotion_analysis

Results

Spam detection

	Accuracy	Precision*	Recall*	F1-score*
LSTM	0.9559	0.9541	0.9545	0.9543
DistilBERT	0.9886	0.9886	0.9877	0.9882

Emotion analysis

	Accuracy	Precision*	Recall*	F1-score*
LSTM	0.8643	0.8242	0.8253	0.8241
DistilBERT	0.9260	0.8912	0.8879	0.8891

* macro-average

License

The code in this project is licensed under the MIT License.