IJCAI 2015 supplementary material
In order to give a qualitative idea of the behavior of our claim detector for argumentation mining, we hereby report the claims extracted by our classifier on 15 Wikipedia pages chosen as follows:
Click on the title of an article in order to see the claims detected by our system (highlighted in
red).
False positives (detected by our system but not labeled as context-dependent claims for a given topic in the IBM corpus) are in
bluewhile false negatives (not detected by our system but labeled as positives in the IBM corpus) are highlighted in
green.