EU Cohesion policy in the media: A computational text analysis of online news, user comments and social media

Abstract

This research paper investigates Cohesion Policy in the mass media by applying computational text analysis to a novel media dataset. Specifically, structural topic modelling and sentiment analysis is applied to online news, user comments and social media at multiple territorial levels. The dataset includes 4,000 news stories, 33,000 user comments, 3,700 posts and 19,500 tweets from Facebook and Twitter respectively, as well as comments and reactions. We discover a two-level hierarchy of descending sentiment on Cohesion policy news stories, whereby international media use more negative sentiment than EU web-native media at one level, and the national media in turn use more negative sentiment than regional level sources at the domestic level. The sentiment of user comments on news articles varies across our country cases, being mainly neutral or positive in Spain and overwhelmingly negative in the United Kingdom even in pro-European news sources. Finally, social media content on Facebook and Twitter is largely neutral, and dominated by official policy channels and stakeholders. We conclude that a territorially-targeted media strategy is needed to improve public appreciation of Cohesion policy, along with more emotive and topical social media activity in order engage and connect with citizens.