PhishGuard Classifier Engine
TF-IDF + XGBoost ensemble trained on mamtakumar/seven-phishing-email-datasets with synthetic Persian phishing samples.
Model Details
| Property | Value |
|---|---|
| Engine | TF-IDF (1024 char n-grams) + XGBoost |
| Reference HF model | aamoshdahal/email-phishing-distilbert-finetuned |
| Training corpus | HF seven-phishing-email-datasets + Persian SE samples |
| Version | 1.0.0 |
Intended Use
- Email phishing triage for SOC / email security teams
- Persian social engineering detection supplement
- Demo and evaluation on synthetic + public corpora
- Not for production email gateway replacement
Limitations
- English-primary training corpus; Persian via pattern rules + synthetic samples
- Header auth analysis is heuristic (not live DNS lookup)
- Requires retraining for organization-specific mail patterns
Built by Aria AI.
- Downloads last month
- 15