Dataset for: "Using AI to detect misinformation and emotions on Telegram: a comparison with the media"

This dataset contains the raw data used in the article “Using AI to detect misinformation and emotions on Telegram: a comparison with the media”, accepted for publication in index.comunicación. The data includes:
 • Telegram dataset (tg_messages.csv): 54,456 posts extracted from 33 public Telegram channels between 23 July and 16 November 2023, related to the political debate around the Amnesty Law in Spain. Each entry includes message metadata such as channel, date, views, and content.
 • News headlines dataset (Titulares.csv): 46,022 news headlines mentioning “amnesty”, extracted from 377 Spanish national media outlets indexed in MediaCloud, during the same period.
 • Analysis scripts: Available upon request or pending publication in the article’s supplementary materials.

The data was used for topic modelling, sentiment and emotion detection with NLP techniques based on Python libraries like BERTopic and pysentimiento. All data is anonymized and publicly accessible or derived from open sources.