Dataset · cardiffnlp
Multilingual X (Twitter) post topic dataset
cardiffnlp/tweet_topic_multilingual
Tweets in four languages labeled across 19 topics for multi-label text classification tasks.
What's inside
The official X-Topic corpus (presented at EMNLP 2024) collected from platform X posts. The card indicates 4,000 tweets—1,000 each in English, Japanese, Spanish, and Greek (metadata notes the 10K-100K range, likely accounting for splits or configs). Texts are labeled across several of 19 thematic categories.
What you can do with it
- Fine-tuning multilingual language models for social media multi-label classification
- Evaluating cross-lingual transfer quality between English, Japanese, Spanish, and Greek
- Creating content filters and trend monitoring systems in X
- Tasks
- Text classification
- Languages
- English, Japanese, Greek, Spanish
- Size
- 10K–100K rows
- Downloads this month
- 2.3K
- Likes
- 4
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("cardiffnlp/tweet_topic_multilingual")