Leveraging Behavioral and Social Information for Weakly Supervised Collective Classification of Political Discourse on Twitter

Abstract

Framing is a political strategy in which politicians carefully word their statements in order to control public perception of issues. Previous works exploring political framing typically analyze frame usage in longer texts, such as congressional speeches. We present a collection of weakly supervised models which harness collective classification to predict the frames used in political discourse on the microblogging platform, Twitter. Our global probabilistic models show that by combining both lexical features of tweets and network-based behavioral features of Twitter, we are able to increase the average, unsupervised F 1 score by 21.52 points over a lexical baseline alone.

Framing #Ferguson: A comparative analysis of media tweets in the U.S., U.K., Spain, and France

“…For example, Boydstun et al (2014) used automated coding and computational analysis to predict the primary frames of news stories based on word patterns. Similarly, Johnson et al (2017) relied on language-based models to determine primary frames of political tweets.…”

Issue Framing in Online Discussion Fora

“…Naderi and Hirst (2017) use the same resource, but make predictions at the sentence level, using topic models and recurrent neural networks. Johnson et al (2017) predict frames in social media data at the micro-post level, using probabilistic soft logic based on lists of keywords, as well as temporal similarity and network structure. All the work mentioned above uses the generic frames of Boydstun et al (2014)'s Policy Frames Codebook.…”

“…We tokenize all sequences using spaCy 7 , which we also use for sentence splitting in the news articles dataset. For the Twitter dataset, we follow Johnson et al (2017) in removing URLs and @-mentions.…”

“…Computational methods for frame classification have previously been studied in news articles (Card et al, 2015) and social media posts (Johnson et al, 2017). In this work, we introduce a new benchmark dataset, based on a subset of the 15 generic frames in the Policy Frames Codebook by Boydstun et al (2014).…”

Mapping (Dis-)Information Flow about the MH17 Plane Crash

“…In order to scale annotations that go beyond metadata to larger datasets, Natural Language Processing (NLP) models can be used to automatically label text content. For example, several works developed classifiers for annotating text content with frame labels that can subsequently be used for large-scale content analysis (Boydstun et al, 2014;Tsur et al, 2015;Card et al, 2015;Johnson et al, 2017;Ji and Smith, 2017;Naderi and Hirst, 2017;Field et al, 2018;Hartmann et al, 2019). Similarly, automatically labeling attitudes expressed in text (Walker et al, 2012;Hasan and Ng, 2013;Augenstein et al, 2016;Zubiaga et al, 2018) can aid the analysis of disinformation and misinformation spread (Zubiaga et al, 2016).…”