fix yaml and add readme
Browse files
README.md
CHANGED
|
@@ -13,4 +13,30 @@ tags:
|
|
| 13 |
pretty_name: pidginshield
|
| 14 |
size_categories:
|
| 15 |
- n<1K
|
| 16 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
pretty_name: pidginshield
|
| 14 |
size_categories:
|
| 15 |
- n<1K
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
# pidginshield: pidgin crypto scam detection dataset
|
| 19 |
+
|
| 20 |
+
## overview
|
| 21 |
+
this dataset is designed to detect crypto scams in nigerian pidgin, english, and mixed-language communication.
|
| 22 |
+
|
| 23 |
+
it captures realistic scam patterns commonly found in west african crypto communities, where informal language and slang are widely used.
|
| 24 |
+
|
| 25 |
+
## dataset structure
|
| 26 |
+
each entry contains:
|
| 27 |
+
- text
|
| 28 |
+
- language_type
|
| 29 |
+
- label (scam or legit)
|
| 30 |
+
- scam_category
|
| 31 |
+
- intent
|
| 32 |
+
- explanation
|
| 33 |
+
|
| 34 |
+
## use cases
|
| 35 |
+
- training scam detection models
|
| 36 |
+
- improving ai safety in fintech and web3
|
| 37 |
+
- moderation tools for crypto communities
|
| 38 |
+
|
| 39 |
+
## attribution
|
| 40 |
+
this dataset, pidginshield, was created for the uncharted data challenge by adaption labs.
|
| 41 |
+
|
| 42 |
+
it leverages adaptive data, adaption’s data infrastructure for building and evolving datasets.
|