How reemaz06/verinews-ai works
VeriNews AI accepts article text, URLs, or documents, then returns a fake-news prediction, calibrated confidence, token-level explanations, and heuristic credibility indicators. Its API also exposes analysis history and analytics backed by SQLite. The graph includes the documented ingestion and inference stages plus distinct service modules present in the tree. Runtime wiring is supported where shown by excerpts or README; connections for unsampled modules are omitted rather than inferred.
Main entry points: main.py FastAPI gateway · explainer.py Text analysis · database.py SQLite access · index.html Web console
Components 17
User interface
- Web console
app/static/index.html
API and ingestion
- FastAPI gateway
app/main.py - URL extraction
app/services/url_extractor.py - File extraction
app/services/file_extractor.py
Analysis services
- Text analysis
ml/explainability/explainer.py - Credibility audit
app/services/credibility.py - Similarity check
app/services/similarity_detector.py - Claims extraction
app/services/claims_extractor.py - Headline analysis
app/services/headline_analyzer.py - Language detection
app/services/language_detector.py - Robustness testing
app/services/robustness_tester.py - XAI evaluation
ml/explainability/run_xai.py - Text preprocessing
ml/preprocessing/preprocess.py
Persistence and ML assets
- Classifier assets
ml/models/best_model.joblib - SQLite access
app/db/database.py - Analysis history
app/data/verinews.db
Other components
- Analyst
Connections 12
- Analyst → Web console: submits article
README.md - Web console → FastAPI gateway: calls REST API
README.md - FastAPI gateway → URL extraction: extracts URL
app/main.py - FastAPI gateway → File extraction: extracts file
app/main.py - FastAPI gateway → Text analysis: requests prediction
app/main.py - FastAPI gateway → Credibility audit: requests scoring
app/main.py - FastAPI gateway → SQLite access: reads and writes
app/main.py - SQLite access → Analysis history: persists records
app/db/database.py - Similarity check → SQLite access: queries history
app/services/similarity_detector.py - Text analysis → Classifier assets: loads model
README.md - Text analysis → Text preprocessing: uses preprocessing
- XAI evaluation → Text analysis: runs explanations
ml/explainability/run_xai.py