Purpose: Automate extraction of crucial information from chat apps for disaster relief.
Inspiration: Based on the work supporting Turkey-Syria earthquake relief operations.
Detailed Report: Check out RelifNER App Detailed Report for a comprehensive overview of the development process.
Custom Text Classifier: Distinguishes between disaster-related and random messages.
Named Entity Recognition (NER): Identifies essential entities like names and phone numbers.
HuggingFace & SpaCy: For training the text classifier and NER model.
Gradio: To create a user-friendly web interface.
Telegram Bot API: For collecting distress messages.
Data Collection: Utilizes synthetic messages generated by ChatGPT.
Model Training: Guides on fine-tuning models using HuggingFace and SpaCy.
Multilingual Support: To cater to diverse linguistic needs.
Model Improvement: Continuous training for enhanced performance.
Feedback: Suggestions for improvements are welcome.
Collaboration: Open to partnerships for further development.