How asantemaicenne/fake_job_detector works
The product is a Flask-based fake-job-ad checker. A user submits pasted or structured job text, an image for OCR, or a public job-post URL; the Flask handler obtains text, classifies it into a risk band, saves the analysis, and returns a result. Separate branches cover OCR and guarded URL fetching, while model training is a supporting product capability. The sampled sources show the handler’s calls and MongoDB write; the README documents the browser workflow and safety limits. Training data and model artifacts are not runtime services.
Main entry points: app.py Flask endpoints · index.html Checker page · classifier.py Risk classifier · train_model.py Training command
Components 11
Web experience
- Checker page
templates/index.html - Flask endpoints
app.py
Text intake
- Text submission
app.py - Image OCR
fake_job_detector/sources.py - URL extraction
fake_job_detector/sources.py
Risk analysis
- Risk classifier
fake_job_detector/classifier.py - Prediction store
fake_job_detector/database.py
Model training
- Labeled jobs CSV
data/sample_jobs.csv - Training command
train_model.py
Other components
- Job checker
- MongoDB
Connections 13
- Job checker → Checker page: uses
README.md - Checker page → Flask endpoints: submits checks
README.md - Flask endpoints → Text submission: validates text
app.py - Flask endpoints → Image OCR: requests OCR
app.py - Flask endpoints → URL extraction: fetches page text
app.py - Flask endpoints → Risk classifier: requests prediction
app.py - Flask endpoints → Prediction store: saves result
app.py - Prediction store → MongoDB: inserts prediction
fake_job_detector/database.py - Flask endpoints → Checker page: returns results
README.md - Job checker → Training command: starts training
README.md - Labeled jobs CSV → Training command: provides examples
README.md - Training command → Risk classifier: trains model
train_model.py - Risk classifier → Flask endpoints: loads model
app.py