Historical Dutch Newspaper Extracts with Contentious Word Annotations

A collection of extracts from historical Dutch newspapers annotated for potentially contentious words. It includes multiple annotations per instance, allowing for the quantification of annotation agreement scores. The data can be used to track semantic change over time.

Use Cases

• Analyze annotation agreement scores to measure consensus on the contentiousness of specific words over time.
• Train a text classification model to identify potentially contentious words in historical Dutch newspaper extracts.
• Track the frequency and contextual usage of annotated contentious words to study semantic shifts.

Strengths

• Contains multiple annotations per instance, enabling analysis of inter-annotator agreement.
• Annotated by both crowdsourced and expert-generated methods, providing varied perspectives.
• Dataset size is categorized as between 1,000 and 10,000 instances.

Limitations

• The specific criteria for labeling words as 'contentious' are not detailed, which may affect reproducibility.
• The dataset's temporal coverage and geographic scope within the Netherlands are unspecified.
• Annotations may reflect present-day sensibilities applied to historical text, introducing potential anachronistic bias.

Provenance

Source

Extracts from historical Dutch newspapers.

Collection Method

Instances were annotated by both crowdsourced and expert-generated methods.

Freshness

Last updated on 2022-08-01.

Geography

Netherlands (based on language and source).