Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
csv
Sub-tasks:
multi-class-classification
Size:
< 1K
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,36 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# Trilingual Fraud & Consumer Protection Dataset (V2)
|
| 6 |
+
|
| 7 |
+
## Overview
|
| 8 |
+
This dataset contains multilingual fraud detection and consumer protection samples across English, Hindi, and Punjabi. It is designed to model real-world scam patterns, including multi-step deception flows, subtle edge cases, and safe advisory responses.
|
| 9 |
+
|
| 10 |
+
## Key Features
|
| 11 |
+
- Multilingual coverage (English, Hindi, Punjabi)
|
| 12 |
+
- Code-mixed conversational data
|
| 13 |
+
- Multi-step fraud sequences (trust-building → payment → urgency)
|
| 14 |
+
- Realistic safe responses and advisory patterns
|
| 15 |
+
- Edge cases reflecting subtle and misleading fraud tactics
|
| 16 |
+
|
| 17 |
+
## Structure
|
| 18 |
+
Columns:
|
| 19 |
+
- text
|
| 20 |
+
- label (scam / safe)
|
| 21 |
+
- category
|
| 22 |
+
- sub_type
|
| 23 |
+
- language
|
| 24 |
+
- code_mixed
|
| 25 |
+
- context_type
|
| 26 |
+
- reasoning_tag
|
| 27 |
+
- difficulty
|
| 28 |
+
|
| 29 |
+
## Use Cases
|
| 30 |
+
- Fraud detection systems
|
| 31 |
+
- LLM safety and alignment research
|
| 32 |
+
- Multilingual NLP evaluation
|
| 33 |
+
- Consumer protection modeling
|
| 34 |
+
|
| 35 |
+
## Notes
|
| 36 |
+
This dataset focuses on realism and diversity rather than synthetic templated generation, making it suitable for advanced evaluation tasks.
|