OWI Argument Extraction Dataset

This dataset contains arguments extracted from web pages collected in the OpenWebIndex OWI (https://dashboard.ows.eu/). The index data were searched for web pages related to 31 controversial topics from the ArgKP dataset. From each web page, argumentative segments were extracted on sentence and paragraph level, applying an argument classification model (github.com/UKPLab/acl2019-BERT-argument-classification). The argumentative segments were then matched to suitable key points - high level argumentative statements that help to summarize the main arguments in a large collection. The key points were also taken from the ArgKP dataset.