Related Experiment Video
Updated: Jun 23, 2025

Training Synesthetic Letter-color Associations by Reading in Color
Published on: February 20, 2014
Dutch tweets before, during, and after a black lives matter demonstration in Amsterdam: Expert annotated data,
Laurens H F Müter1,2, Christof van Nimwegen2, Remco C Veltkamp2
1Netherlands Police, Van Deventerlaan 50, 3528 AE, Utrecht, the Netherlands.
Abstract:
The Netherlands police are looking for measures to examine sentiment on social media related to protest demonstrations. While models exist to detect more subtle expressions of sentiment within tweets, models trained in the Dutch language are scarce. Being able to predict sentiment development during protests is relevant for parties like the Dutch government and the police to get more insight to when and where potential law enforcement is needed for public order and safety. Therefore, to analyse sentiment before, during, and after protest demonstrations, data was collected with tweets related to a Black Lives Matter protest that took place in Amsterdam during the COVID-19 pandemic. All tweets have been manually labelled by a dedicated open-source intelligence (OSINT) team within the Netherlands police following an established protocol. Both the data and the protocol are available, and interesting for researchers in natural language processing, topic detection, sentiment analysis, and protests analysis. The developed labelling tool for the labelling process is publicly available.
Related Concept Videos
Milgram's Obedience to Authority
Archival Research
Relationship Formation

