Agreement and Disagreement: Comparison of Points of View in the Political Domain

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 2461–2470, Osaka, Japan, December 11-17 2016. Agreement and Disagreement: Comparison of Points of View in the Political Domain Stefano Menini University of Trento Fondazione Bruno Kessler Trento, Italy menini@fbk.eu Sara Tonelli Fondazione Bruno Kessler Trento, Italy satonelli@fbk.eu Abstract The automated comparison of points of view between two politicians is a very challenging task, due not only to the lack of annotated resources, but also to the different dimensions participating to the definition of agreement and disagreement. In order to shed light on this complex task, we first carry out a pilot study to manually annotate the components involved in detecting agreement and disagreement. Then, based on these findings, we implement different features to capture them automatically via supervised classification. We do not focus on debates in dialogical form, but we rather consider sets of documents, in which politicians may express their position with respect to different topics in an implicit or explicit way, like during an electoral campaign. We create and make available three different datasets. 1 Introduction When it comes to evaluate whether the statements of two persons are in agreement or not about a topic, several past works approach the problem by classifying the single statements as supporting or opposing the topic, considering the task as a variant of sentiment analysis (Somasundaran and Wiebe, 2010). These approaches proved to be reliable in specific settings, where the goal of the statements was to express support or opposition w.r.t. the topic. However, when applied to the political domain, they often result into an oversimplified representation of the dynamics involved in the comparison of two positions. In our view, several aspects contribute to the assessment of agreement and disagreement in the political domain, requiring to be properly addressed. As an example, let us consider two excerpts uttered by Kennedy and Nixon in 1960 about the situation in Cuba under the Castro regime: Kennedy: “There is not any doubt we had great influence in Cuba, and I think it is unfortu- nate that we did not use that influence more vigorously to persuade Castro to hold free, open elections, so that the people of Cuba could have made the choice.” Nixon: “What we must remember too is that the United States has the military power - and Mr. Castro knows this - to throw him out of office tomorrow or the next day or any day that we choose.” The two examples show that neither Nixon nor Kennedy support the Castro regime and that they both share the same negative sentiment about it in their speeches. But beside being on the same side in contrasting the regime, their points of view on it are, from a pragmatic perspective, very different: while Kennedy supports free elections, Nixon does not hesitate to remark United States military power and the possibility to remove Castro in any moment. Overall, the two positions are in disagreement. This example shows that the sentiment and …

Content truncated to 500 words