The second part of the arg-microtexts corpus contains 171 additional short argumentative texts in English, annotated with argumentation structure.
For a list of the topics and trigger questions, see topics_triggers.md.
One of the features new in this part of the corpus are implicit main claims
(see Section 4.1 of our paper). Note that discourse units that have been
introduced to represent implicit claims are marked in the xml-files with the
implicit="true"
property. In the source text, they are highlighted as fully
uppercase segments (which do not exist otherwise in the corpus). See e.g. this
xml-file and source text.
The arg-microtexts-part2 corpus is released under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. You can find a human-readable summary of the licence agreement here:
https://creativecommons.org/licenses/by-nc-sa/4.0/
If you are using our corpus for research purposes, please cite the following paper:
Maria Skeppstedt, Andreas Peldszus, Manfred Stede. More or less controlled elicitation of argumentative text: Enlarging a microtext corpus via crowdsourcing. In: Proceedings of the 5th Workshop on Argument Mining. EMNLP 2018, Belgium, Brussels, November 2018
arg-microtexts - First part of the corpus.