AINS 2025 hackathon winner (2nd) team solution. An-AI powered tool that detects scams (even in native Tunisian) and provides explanations.
# 🛡️ MYTH CHASER
**Anti-scam and myth-busting utility powered by AI**
MYTH CHASER is an advanced AI-powered fact-checking and anti-scam detection system that helps users verify the authenticity of claims, statements, and media content. Using multiple machine learning models and real-time web search capabilities, it provides comprehensive analysis to identify facts, myths, and potential scams.
## ✨ Features
### 🤖 Multi-Model AI Analysis
- **Natural Language Inference (NLI)**: RoBERTa-based model for logical reasoning between claims and evidence
- **Sentence-BERT (SBERT)**: Semantic similarity analysis using sentence embeddings
- **ClaimBuster Integration**: Professional fact-checking API for claim verification
- **Google Fact Check API**: Access to Google's comprehensive fact-checking database
- **Fake News Detection**: Specialized model for identifying misinformation patterns
- **Groq Qwen3-32B**: Advanced large language model with sophisticated reasoning capabilities
- **TunBERT**: Specialized Arabic and Tunisian dialect fact-checking model
- **AI Explanation Generator**: Groq-powered detailed explanations for all verdicts
### 🔍 Multi-Format Content Analysis
- **Text Processing**: Direct text input analysis and verification
- **Image Analysis**: OCR text extraction and visual content description using BLIP
- **Audio Processing**: Speech-to-text conversion for audio content verification
- **Drag & Drop Interface**: Seamless file upload with support for multiple formats
### 🌐 Multi-Language Support
- **Auto-Detection**: Automatic language identification for incoming content
- **Translation Engine**: Google Translate integration for seamless cross-language analysis
- **Arabic Dialect Support**: Specialized handling for Tunisian Arabic and transliterated text
- **Language Preservation**: Original text maintained for dialect-specific models
### 🌐 Real-Time Web Search
- **DuckDuckGo Integration**: Automated web search for evidence gathering
- **Source Aggregation* …