In Plain Sight: Media Bias Through the Lens of Factual Reporting

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 6343–6349, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 6343 In Plain Sight: Media Bias Through the Lens of Factual Reporting Lisa Fan1,∗ Marshall White1,∗ Eva Sharma1 Ruisi Su1 Prafulla Kumar Choubey2 Ruihong Huang2 Lu Wang1 1 Khoury College of Computer Sciences, Northeastern University, Boston, MA 02115 2 Department of Computer Science and Engineering, Texas A&M University {fan.lis, white.mars, sharma.ev, su.ruis}@husky.neu.edu {prafulla.choubey, huangrh}@tamu.edu, luwang@ccs.neu.edu Abstract The increasing prevalence of political bias in news media calls for greater public awareness of it, as well as robust methods for its detec- tion. While prior work in NLP has primarily focused on the lexical bias captured by lin- guistic attributes such as word choice and syn- tax, other types of bias stem from the actual content selected for inclusion in the text. In this work, we investigate the effects of infor- mational bias: factual content that can never- theless be deployed to sway reader opinion. We first produce a new dataset, BASIL, of 300 news articles annotated with 1,727 bias spans1 and find evidence that informational bias appears in news articles more frequently than lexical bias. We further study our an- notations to observe how informational bias surfaces in news articles by different media outlets. Lastly, a baseline model for infor- mational bias prediction is presented by fine- tuning BERT on our labeled data, indicating the challenges of the task and future directions. 1 Introduction News media exercises the vast power of swaying public opinion through the way it selects and crafts information (De Vreese, 2004; DellaVigna and Gentzkow, 2010; McCombs and Reynolds, 2009; Perse, 2001; Reynolds and McCombs, 2002). Multiple studies have identified the correlation be- tween the increasing polarization of media and the general population’s political stance (Gentzkow and Shapiro, 2010, 2011; Prior, 2013), underscor- ing the imperative to understand the nature of news bias and how to accurately detect it. In the natural language processing community, the study of bias has centered around what we term ∗Equal contribution. Lisa Fan focused on annotation schema design and writing, Marshall White focused on data collection and statistical analysis. 1Dataset can be found at www.ccs.neu.edu/home/ luwang/data.html. Main event: Democratic presidential candidates ask to see full Mueller report Main targets: Donald Trump, Democratic candidates HPO: Democrats want access to special counsel Robert Mueller’s investigation into Russian interference in the 2016 presidential election [before President Don- ald Trump has a chance to interfere.]Trump ... Sen. Mark Warner said in a statement: [“Any attempt by the Trump Administration to cover up the results of this investigation into Russia’s attack on our democracy would be unacceptable.”]Trump FOX: Democratic presidential candidates [wasted no time]Dems Friday evening demanding the immediate public release of the long-awaited report from Robert S. Mueller III. ... Several candidates, in calling for the swift release of the report, also [sought to gather new supporters and their email addresses]Dems …

Content truncated to 500 words