Framing Unpacked: A Semi-Supervised Interpretable Multi-View Model of Media Frames

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2154–2166 June 6–11, 2021. ©2021 Association for Computational Linguistics 2154 Framing Unpacked: A Semi-Supervised Interpretable Multi-View Model of Media Frames Shima Khanehzar† Trevor Cohn† Gosia Mikolajczak� Andrew Turpin† Lea Frermann† † School of Computing and Information Systems �School of Social and Political Sciences The University of Melbourne skhanehzar@student.unimelb.edu.au {tcohn,mmikolajcza,aturpin,lfrermann}@unimelb.edu.au Abstract Understanding how news media frame politi- cal issues is important due to its impact on public attitudes, yet hard to automate. Com- putational approaches have largely focused on classifying the frame of a full news article while framing signals are often subtle and local. Furthermore, automatic news analy- sis is a sensitive domain, and existing clas- sifiers lack transparency in their predictions. This paper addresses both issues with a novel semi-supervised model, which jointly learns to embed local information about the events and related actors in a news article through an auto-encoding framework, and to leverage this signal for document-level frame classifica- tion. Our experiments show that: our model outperforms previous models of frame predic- tion; we can further improve performance with unlabeled training data leveraging the semi- supervised nature of our model; and the learnt event and actor embeddings intuitively corrob- orate the document-level predictions, provid- ing a nuanced and interpretable article frame representation. 1 Introduction Journalists often aim to package complex real- world events into comprehensive narratives, follow- ing a logical sequence of events involving a limited set of actors. Constrained by word limits, they nec- essarily select some facts over others, and make cer- tain perspectives more salient. This phenomenon of framing, be it purposeful or unconscious, has been thoroughly studied in the social and political sciences (Chong and Druckman, 2007). More re- cently, the natural language processing community has taken an interest in automatically predicting the frames of news articles (Card et al., 2016; Field et al., 2018; Akyürek et al., 2020; Khanehzar et al., 2019; Liu et al., 2019a; Huguet Cabot et al., 2020). Definitions of framing vary widely including: expressing the same semantics in different forms (equivalence framing); presenting selective facts and aspects (emphasis framing); and using estab- lished syntactic and narrative structures to convey information (story framing) (Hallahan, 1999). The model presented in this work builds on the concepts of emphasis framing and story framing, predicting the global (aka. primary) frame of a news article on the basis of the events and participants it features. Primary frame prediction has attracted substan- tial interest recently with the most accurate models being supervised classifiers built on top of large pre- trained language models (Khanehzar et al., 2019; Huguet Cabot et al., 2020). This work advances prior work in two ways. First, we explicitly incor- porate a formalization of story framing into our frame prediction models. By explicitly modeling news stories as latent representations over events and related actors, we obtain interpretable, latent representations lending transparency to our frame prediction models. We argue that transparent ma- …

Content truncated to 500 words