bolewara/hinglish-scam-text-dataset

Hinglish & English Financial Scam/Fraud Text Dataset

Curated dataset for detecting financial fraud, phishing, and scam messages in English and Hinglish (code-mixed Hindi + English). Created to fine-tune bolewara/financial-fraud-scam-detector.

Contents — 3,787 examples

Split	Examples	Source
English benign	~3,000	SMS Spam Collection (ham) + curated
English scam	~700	SMS Spam Collection (spam) + curated phishing/fraud texts
Hinglish scam	35	Hand-written realistic Indian fraud patterns (KYC, UPI, lottery, loan, OTP, PAN, electricity scams)
Hinglish benign	28	Hand-written normal Hindi/Hinglish messages

Labels

• 0 — Benign / legitimate
• 1 — Scam / fraudulent

Hand-written Hinglish scam patterns covered

• Bank KYC expiry / Aadhaar link blocking
• UPI fraud / lottery winnings / prize claims
• Fake loan offers with upfront fees
• PAN suspension / tax refund phishing
• Fake job offers with registration fees
• Subscription/supply suspension threats
• Emergency money fraud ("your son needs money")
• Cashback refund phishing
• OTP harvesting

Format

JSON file with rows: [{"text": "...", "label": 0|1}, ...]

Provenance

Curated by Anuj Bolewar. English portions derived from the public SMS Spam Collection (UCI). Hinglish portions are original, written to mirror real scam patterns reported by Indian banks and cyber cells. For research and education.