From legal to technical concept: Towards an automated classification of German political Twitter postings as criminal offenses

Proceedings of NAACL-HLT 2019, pages 1337–1347 Minneapolis, Minnesota, June 2 - June 7, 2019. c⃝2019 Association for Computational Linguistics 1337 From Legal to Technical Concept: Towards an Automated Classification of German Political Twitter Postings as Criminal Offenses Frederike Zufall1†, Tobias Horsmann2†, Torsten Zesch2 1 Waseda Institute for Advanced Study, Waseda University, Tokyo, Japan 2 Language Technology Lab, University of Duisburg-Essen, Germany f.zufall@kurenai.waseda.jp {tobias.horsmann|torsten.zesch}@uni-due.de Abstract Advances in the automated detection of of- fensive Internet postings make this mechanism very attractive to social media companies, who are increasingly under pressure to monitor and action activity on their sites. However, these advances also have important implications as a threat to the fundamental right of free expres- sion. In this article, we analyze which Twitter posts could actually be deemed offenses un- der German criminal law. German law follows the deductive method of the Roman law tradi- tion based on abstract rules as opposed to the inductive reasoning in Anglo-American com- mon law systems. This allows us to show how legal conclusions can be reached and imple- mented without relying on existing court deci- sions. We present a data annotation schema, consisting of a series of binary decisions, for determining whether a specific post would constitute a criminal offense. This schema serves as a step towards an inexpensive cre- ation of a sufficient amount of data for auto- mated classification. We find that the majority of posts deemed morally offensive actually do not constitute a criminal offense and still con- tribute to public discourse. Furthermore, lay- men can provide sufficiently reliable data to an expert reference but are, for instance, more le- nient in the interpretation of what constitutes a disparaging statement. 1 Introduction The Internet is frequently used for discussing a va- riety of topics and an important medium for the ex- change of opinions, considered crucial for healthy democratic societies. However, the rough tone in the Internet frequently leads to defamatory or abusive comments in these discussions. The EU has tried to tackle the problem by defining the †Equal contribution. term ‘illegal hate speech’.1 Additionally, in 2017, the European Commission published a communi- cation entitled ‘Tackling Illegal Content Online’ aiming for enhanced responsibility of online plat- forms.2 Independently from these recent devel- opments on the EU level, Germany adopted the ‘Network Enforcement Act’3 in 2017. The Act provides for a regulatory framework for ‘illegal content’4 on social network platforms like Twitter or Facebook. It imposes the obligation on these providers to delete illegal content upon notifica- tion within seven days; in case of evidently illegal content within 24 hours.5 From a practical point of view, given the number of statements on social media along with their possible notification, fea- sibility and accuracy of the required legal assess- ment becomes an important issue. Natural Lan- guage Processing might thus provide the neces- sary means to assist the legal assessment. In this work, we investigate at which point morally offensive statements in social media con- stitute defamatory offenses under the German Criminal Code (StGB)6, …

Content truncated to 500 words