How keerthanasathish985-beep/ghostsignal works
GhostSignal is an investigation and text-analysis application. Its visible API workflow accepts a topic, obtains a classifier and investigation service, and returns collected records with sentiment statistics and a timeline; it also exposes single-text and batch classification. The classifier loads separate sentiment and authenticity pipelines. The tree documents broader collection, analysis, and reporting capabilities, but most of those modules were not sampled, so their runtime connections are not asserted. The frontend entry point renders the main app; its API behavior was not inspected.
Main entry points: train.py Model training · investigation_service.py Investigation service · classifier.py Prediction classifier · app.py Flask API
Components 22
User interface and API
- Web application
frontend/src/App.tsx - App entry point
frontend/src/main.tsx - Flask API
app.py
Investigation and collection
- Investigation service
src/services/investigation_service.py - GDELT news collection
src/collectors/news/gdelt_collector.py - Discussion collection
src/collectors/discussions - Other source collectors
src/collectors
Analysis and reporting
- Text preprocessing
src/preprocessing - Sentiment analysis
src/analysis/engagement.py - Coordination analysis
src/analysis/coordination.py - Similarity analysis
src/analysis/similarity.py - Timeline analysis
src/analysis/timeline.py - Report generation
src/reporting/report_generator.py
Classification models
- Prediction classifier
src/model/classifier.py - Model training
src/model/train.py - Model evaluation
src/model/evaluate.py - Training dataset
data/training/training_data.csv - Sentiment pipeline
models/sentiment_pipeline.joblib - Authenticity pipeline
models/authenticity_pipeline.joblib
Other components
- User
- GDELT API
- Hacker News
Connections 13
- User → Web application: uses
- App entry point → Web application: renders
frontend/src/main.tsx - Flask API → Prediction classifier: predicts with
app.py - Flask API → Investigation service: invokes
app.py - Investigation service → GDELT news collection: collects through
- Investigation service → Discussion collection: collects through
- GDELT news collection → GDELT API: requests articles
src/collectors/news/gdelt_collector.py - Model training → Training dataset: loads data
- Model training → Model evaluation: evaluates with
src/model/train.py - Model training → Sentiment pipeline: writes pipeline
src/model/train.py - Model training → Authenticity pipeline: writes pipeline
src/model/train.py - Prediction classifier → Sentiment pipeline: loads pipeline
src/model/classifier.py - Prediction classifier → Authenticity pipeline: loads pipeline
src/model/classifier.py