Neural Based Statement Classification for Biased Language

Abstract

Biased language commonly occurs around topics which are of controversial nature, thus, stirring disagreement between the different involved parties of a discussion. This is due to the fact that for language and its use, specifically, the understanding and use of phrases, the stances are cohesive within the particular groups. However, such cohesiveness does not hold across groups.In collaborative environments or environments where impartial language is desired (e.g. Wikipedia, news media), statements and the language therein should represent equally the involved parties and be neutrally phrased. Biased language is introduced through the presence of inflammatory words or phrases, or statements that may be incorrect or one-sided, thus violating such consensus.In this work, we focus on the specific case of phrasing bias, which may be introduced through specific inflammatory words or phrases in a statement. For this purpose, we propose an approach that relies on a recurrent neural networks in order to capture the inter-dependencies between words in a phrase that introduced bias.We perform a thorough experimental evaluation, where we show the advantages of a neural based approach over competitors that rely on word lexicons and other hand-crafted features in detecting biased language. We are able to distinguish biased statements with a precision of P = 0.917, thus significantly outperforming baseline models with an improvement of over 30%. Finally, we release the largest corpus of statements annotated for biased language.

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

“…1 Dataset can be found at www.ccs.neu.edu/home/ luwang/data.html. lexical bias: bias stemming from content realization, or how things are said (Greene and Resnik, 2009;Hube and Fetahu, 2019;Iyyer et al, 2014;Recasens et al, 2013;Yano et al, 2010). Such forms of bias typically do not depend on context outside of the sentence and can be alleviated while maintaining its semantics: polarized words can be removed or replaced, and clauses written in active voice can be rewritten in passive voice.…”

Automatically Neutralizing Subjective Bias in Text

“…The study of subjectivity in NLP was pioneered by the late Janyce Wiebe and colleagues (Bruce and Wiebe 1999;Hatzivassiloglou and Wiebe 2000). Several studies develop methods for highlighting subjective or persuasive frames in a text (Rashkin et al 2017;Tsur, Calacci, and Lazer 2015), or detecting biased sentences (Hube and Fetahu 2018;Morstatter et al 2018;Yang et al 2017;Hube and Fetahu 2019) of which the most similar to ours is Recasens, Danescu-Niculescu-Mizil, and Jurafsky (2013), whose early, smaller version of WNC and logistic regression-based bias detector inspired our study.…”

Computational Ad Hominem Detection

“…Crowdsourcing the labeling is another option; either by paying the participants (Hube and Fetahu, 2018) or by providing a service in return, like a game (Habernal et al, 2017).…”

“…For verbs, the tags can also encode the tense, and further information can be contained. These labels have been used successfully in NLP tasks (Hube and Fetahu, 2018). In this work, the POS tagging is done by the Python library NLTK (Bird et al, 2009), which uses the Penn Treebank tagset.…”