---
task_categories:
- text-classification
- image-classification
language:
- el
size_categories:
- n<1K
---

This is a dataset for sentiment analysis created by text-image pairs collected from greek Twitter.  
posted: from April 2023 to September 2023  
Context: general purpose, mostly politics and athletics  
Total pairs: 260  
The purpose of the dataset is to be used as a test dataset, not for training/fine-tuning a model.  
Labels: Negative, Neutral, Positive  
labelling.xlsx contains the labels for texts,images and both modalities.