About
Methodology
Dataset
Employment Scam Aegean Dataset (EMSCAD). HTML stripped, t/f labels mapped, no silent text imputation.
Model
TF-IDF + class-weighted logistic regression, selected by validation PR-AUC — not an LLM.
Signals
Fraud score, salary anomaly, duplicates, and company completeness — reported separately.
Limits
Predictions may be wrong. Never definitive accusations. Verify employers independently.