The Metalogue Debate Trainee Corpus: Data Collection and Annotations

The Metalogue Debate Trainee Corpus: Data Collection and Annotations Volha Petukhova1, Andrei Malchanau1, Youssef Oualil1, Dietrich Klakow1, Saturnino Luz2, Fasih Haider3, Nick Campbell3, Dimitris Koryzis4, Dimitris Spiliotopoulos5, Pierre Albert2, Nicklas Linz6, Jan Alexandersson6 1Saarland University, Spoken Language Systems, Saarbr¨ucken, Germany 2The University of Edinburgh, UK; 3Trinity College Dublin, Ireland 4Hellenic Parliament, Greece; 5University of Houston, TX, USA 6German Research Center for Artificial Intelligence (DFKI), Germany 1{v.petukhova,andrei.malchanau,youssef.oualil,dietrich.klakow}@lsv.uni-saarland.de; 2{s.luz,pierre.albert}@ed.ac.uk;3{haiderf,nick}@tcd.ie; 4dkoryzis@parliament.gr 5dspiliotopoulos@uh.edu; 6{nicklas.linz,jan.alexandersson}@dfki.de Abstract This paper describes the Metalogue Debate Trainee Corpus (DTC). DTC has been collected and annotated in order to facilitate the design of instructional and interactive models for Virtual Debate Coach application - an intelligent tutoring system used by young parliamentarians to train their debate skills. The training is concerned with the use of appropriate multimodal rhetorical devices in order to improve (1) the organization of arguments, (2) arguments’ content selection, and (3) argument delivery techniques. DTC contains tracking data from motion and speech capturing devices and semantic annotations - dialogue acts - as defined in ISO 24617-2 and discourse relations as defined in ISO 24617-8. The corpus comes with a manual describing the data collection process, annotation activities including an overview of basic concepts and their definitions including annotation schemes and guidelines on how to apply them, tools and other resources. DTC will be released in the ELRA catalogue in second half of 2018. Keywords: debate argumentation, multimodal data collection, ISO standard annotations 1. Introduction We currently observe a steadily growing interest of re- searchers and practitioners in natural argumentation mod- elling and in developing argumentation technologies. There are systems developed and deployed for legal domains to assist the lawyer in his search for similar past cases, (Teufel, 1999; Br¨uninghaus and Ashley, 2005); for mining argu- ments in social media with the goal to predict consumers sentiment (Bai, 2011), to analyse opinions in public discus- sions (Murakami and Raymond, 2010), to study citizen en- gagement (Purpura et al., 2008) and to recognize stance in political online debates (Somasundaran and Wiebe, 2010; Walker et al., 2012a). Argumentation constitutes an important component of hu- man intelligence. Educational studies have shown that con- structing arguments and engaging in argumentative discus- sion enhance conceptual understanding of the subject mat- ter (Wiley and Voss, 1999; Zohar and Nemet, 2002). Ar- gumentation training systems are designed for the legal do- mains, e.g. to training hypothetical reasoning (Ashley et al., 2007). The TruthMapping1 web application facilitates col- laborative learning through argumentation. DebateGraph2 used to train how to prevent opinion manipulation marking inconsistent arguments. These and other developments were supported by cor- pora collected for various genres, domains and modali- ties. For example, the AIFdb corpora collection (Lawrence and Reed, 2014) of the Centre for Argument Technology, University of Dundee includes data harvested and anal- ysed from ArguBlogging3, BBC Radio programmes (e.g. 1https://www.truthmapping.com/ 2http://debategraph.org/ 3http://www.argublogging.com/ MM2012), Araucaria argument database (Reed, 2006). There is the Internet Argument Corpus (IAC) (Walker et al., 2012b) of political debates on internet forums, consisting of about 11,000 discussions and 390,000 posts. Subsets …

Content truncated to 500 words