Skip to content

Dataset · FredZhang7

Multilingual spam and scam email corpus

FredZhang7/all-scam-spam

Over 42 thousand emails and messages in 43 languages with binary spam labels for training anti-spam filters.

What's inside

Contains 42,619 preprocessed text messages and emails in 43 languages labeled for spam (is_spam=1 or 0). Part of the data, consisting of 1,040 balanced rows (regular correspondence and scam emails in roughly 10 languages), was manually collected and labeled using ChatGPT, while the rest was processed using scripts based on SpamAssassin and Enron corpora.

What you can do with it

  • Training multilingual spam and phishing classifiers
  • Evaluating models in zero-shot classification mode on low-resource languages
  • Filtering incoming messages and emails in mail clients and messengers
Tasks
Text classification, Zero shot classification
Languages
Norwegian, Spanish, Somali, Catalan, Afrikaans, Italian, Dutch, Hindi and 4 more
Size
10K–100K rows
Downloads this month
511
Likes
16

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("FredZhang7/all-scam-spam")