Dutch Twitter Dataset for Rumour Detection, April 2022

A collection of Dutch tweets and features gathered in April 2022 using the Twitter API. A small portion of the tweets are annotated by volunteers for the main task of identifying rumours. The dataset was contributed by Nicky van der Linden and is licensed under CC0-1.0.

Use Cases

• Train a classifier to detect rumours based on features like account verification and follower ratio.
• Analyze the relationship between tweet characteristics (e.g., question marks, capital letters) and rumour status.
• Study rumour propagation within specific Dutch Twitter hashtags (#jinek, #vleestaks, #inflatie).
• Investigate account-level features (like account age and activity) as predictors for rumour spread.

Strengths

• Includes a mix of annotated (rumour/non-rumour) and unannotated tweets for semi-supervised learning.
• Features engineered for rumour detection, such as 'followers_ratio', 'capital letters', and 'time_after_posting'.
• Data is focused on three specific Dutch hashtags, providing a controlled thematic scope.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Twitter API

Collection Method

Gathered via the Twitter API; a portion annotated by volunteer annotators.

Time Range

April 2022

Freshness

Data was gathered in April 2022; last update date is unknown.

Geography

Likely Netherlands or Dutch-speaking region, based on language and hashtags.