Topic-Based Agreement and Disagreement in US Electoral Manifestos

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2938–2944 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Topic-Based Agreement and Disagreement in US Electoral Manifestos Stefano Menini1,3, Federico Nanni2, Simone Paolo Ponzetto2, Sara Tonelli1 1Fondazione Bruno Kessler, Trento, Italy 2University of Mannheim, Germany 3University of Trento, Italy {menini, satonelli}@fbk.eu {federico, simone}@informatik.uni-mannheim.de Abstract We present a topic-based analysis of agreement and disagreement in political manifestos, which relies on a new method for topic detection based on key con- cept clustering. Our approach outperforms both standard techniques like LDA and a state-of-the-art graph-based method, and provides promising initial results for this new task in computational social science. 1 Introduction During the last decade, the adoption of natu- ral language processing (NLP) techniques for the study of political phenomena has gained consid- erable momentum (Grimmer and Stewart, 2013), arguably because of both the availability of parlia- mentary proceedings (van Aggelen et al., 2017), electoral manifestos (Volkens et al., 2011) and campaign debates (Woolley and Peters, 2008), and the interest of the computational social science (CSS) community in the potential of text mining methods for advancing political science research (Lazer et al., 2009). Previous work focused on the automatic de- tection of sentiment expressions in political news (Young and Soroka, 2012), the identification of ideological proportions (Sim et al., 2013) and the scaling on a left-right spectrum of politicians’ speeches (Slapin and Proksch, 2008). More re- cently, researchers looked at topic-centered ap- proaches to provide finer-grained analyses, includ- ing segmentation methods for topic-labeled man- ifestos (Glavaˇs et al., 2016), supporting manual coders in identifying coarse-grained political top- ics (Zirn et al., 2016), as well as topic-based and cross-lingual political scaling (Nanni et al., 2016; Glavaˇs et al., 2017). Measuring Agreement. Automatically measur- ing the level of agreement in political documents (Gottipati et al., 2013; Menini and Tonelli, 2016) has the potential of supporting political analyses such as the comparisons between campaign strate- gies (Burton et al., 2015), the study of promises kept and broken after elections (Naurin, 2011), the formation of coalitions (Debus, 2009) and the interactions between government and opposition (Hix and Noury, 2016). However, previous work relies on the availability of pre-defined topics, in- cluding supervised methods (Galley et al., 2004; Hillard et al., 2003), approaches leveraging col- laboratively generated resources (Gottipati et al., 2013; Awadallah et al., 2012) or pairwise agree- ment detection from political debates (Menini and Tonelli, 2016). Our Contributions. a) New task: Given a collec- tion of political documents such as, e.g., electoral manifestos, we look at ways to perform an auto- matic, topic-based agreement-disagreement clas- sification. b) New approach: We first segment the texts into coarse-grained domains. Next, coarse domains are used to extract a fine-grained list of topic-based points of view which, in turn, are used to perform classification. We achieve this by de- veloping a novel approach for topic detection on the basis of key concept clustering techniques: this is shown to outperform not only LDA-based Topic Modeling …

Content truncated to 500 words