Come hither or go away? Recognising pre-electoral coalition signals in the news

Rehbein, Ines ; Ponzetto, Simone Paolo ; Adendorf, Anna ; Bahnsen, Oke ; Stoetzer, Lukas F. ; Stuckenschmidt, Heiner

URL:	https://aclanthology.org/2021.emnlp-main.615
Document Type:	Conference or workshop publication
Year of publication:	2021
Book title:	Proceedings of the 2021 conference on empirical methods in natural language processing (EMNLP 2021)
Page range:	7798-7810
Conference title:	EMNLP 2021
Location of the conference venue:	Punta Cana, Dominican Republic, online
Date of the conference:	7.-11.11.2021
Publisher:	Moens, Marie-Francine ; Huang, Xuanjing ; Specia, Lucia ; Yih, Scott Wen-tau
Place of publication:	Punta Cana, Dominican Republic
Publishing house:	Association for Computational Linguistics
Publication language:	English
Institution:	Außerfakultäre Einrichtungen > SFB 884 Außerfakultäre Einrichtungen > Graduate School of Economic and Social Sciences- CDSS (Social Sciences)
Subject:	004 Computer science, internet
Keywords (English):	political text analysis
Abstract:	In this paper, we introduce the task of political coalition signal prediction from text, that is, the task of recognizing from the news coverage leading up to an election the(un)willingness of political parties to form a government coalition. We decompose our problem into two related, but distinct tasks: (i) predicting whether a reported statement from a politician or a journalist refers to a potential coalition and (ii) predicting the polarity of the signal — namely, whether the speaker is in favour of or against the coalition. For this, we explore the benefits of multi-task learning and investigate which setup and task formulation is best suited for each subtask. We evaluate our approach, based on hand-coded newspaper articles, covering elections in three countries (Ireland, Germany, Austria) and two languages (English, German). Our results show that the multi-task learning approach can further improve results over a strong monolingual transfer learning baseline.