COVID-19 Disinformation Multilingual Tweets in Four Languages

QCRI's COVID-19 Infodemic Multilingual Dataset contains tweets annotated with fine-grained labels. The data includes tweets in Arabic, Bulgarian, Dutch, and English. The dataset was last updated on September 9, 2024.

Use Cases

• Binary classification for misinformation detection based on annotated tweet content.
• Multiclass classification for categorizing different types of misinformation mentioned in the description.
• Cross-lingual analysis of misinformation narratives based on tweets in Arabic, Bulgarian, Dutch, and English.
• Training models for automated fact-checking systems based on labeled social media data.

Strengths

• Dataset is annotated with fine-grained labels for detailed analysis.
• Multilingual content includes tweets in Arabic, Bulgarian, Dutch, and English.
• Curated to address questions from journalists, fact-checkers, and policymakers.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic and temporal bias inherent to the tweet collection.

Provenance

Source

QCRI

Collection Method

Likely collected from social media platforms, based on the description of containing tweets.

Freshness

Last updated 2024-09-09 06:58:27; freshness should be verified.