Detecting political biases of named entities and hashtags on Twitter

Abstract

Ideological divisions in the United States have become increasingly prominent in daily communication. Accordingly, there has been much research on political polarization, including many recent efforts that take a computational perspective. By detecting political biases in a text document, one can attempt to discern and describe its polarity. Intuitively, the named entities (i.e., the nouns and the phrases that act as nouns) and hashtags in text often carry information about political views. For example, people who use the term “pro-choice” are likely to be liberal and people who use the term “pro-life” are likely to be conservative. In this paper, we seek to reveal political polarities in social-media text data and to quantify these polarities by explicitly assigning a polarity score to entities and hashtags. Although this idea is straightforward, it is difficult to perform such inference in a trustworthy quantitative way. Key challenges include the small number of known labels, the continuous spectrum of political views, and the preservation of both a polarity score and a polarity-neutral semantic meaning in an embedding vector of words. To attempt to overcome these challenges, we propose the Polarity-aware Embedding Multi-task learning (PEM) model. This model consists of (1) a self-supervised context-preservation task, (2) an attention-based tweet-level polarity-inference task, and (3) an adversarial learning task that promotes independence between an embedding’s polarity component and its semantic component. Our experimental results demonstrate that our PEM model can successfully learn polarity-aware embeddings that perform well at tweet-level and account-level classification tasks. We examine a variety of applications—including a study of spatial and temporal distributions of polarities and a comparison between tweets from Twitter and posts from Parler—and we thereby demonstrate the effectiveness of our PEM model. We also discuss important limitations of our work and encourage caution when applying the PEM model to real-world scenarios.

Evolving linguistic divergence on polarizing social media

“…Previous research using social media data to examine political bias has used various strategies to assign a political category to users. Some research uses self-identification, for example by focusing on prominent individuals or smaller samples of prolific public figures with already known political affiliation (Chin et al 2022;Penelas-Leguía et al 2023;Wignell et al 2020;Xiao et al 2022), or collecting data from defined subsections of platforms or discussion forums as the niches or samples of interest (Altmann et al 2011;Soliman et al 2019;Stewart and Eisenstein 2018). Other approaches rely on user characteristics or behavior, using geographical region where geolocation is available (Louf et al 2023a, Louf et al 2023b), sampling data by topically relevant keywords or hashtags (Chen et al 2021;Demszky et al 2019;Oakey et al 2022), categorizing usergenerated content (Fraxanet et al 2023) or clustering networks built from retweeting/reposting or follower data (Conover et al 2011).…”

Online Social Behavior Enhanced Detection of Political Stances in Tweets

“…Ng and Carley (2022) explored several crucial aspects related to cross-dataset model generalization in stance detection, providing practical guidance on effectively generalizing models to new data in real-world applications. Xiao et al (2023) introduced a method to quantify political polarities in tweets by assigning polarity scores to entities and hashtags.…”

“…In this study, we specifically focus on the 2020 US presidential election, intending to ascertain the political polarities of election-related tweets posted during the pre-election period. Unlike news articles Goldwasser 2019, 2021b;, debate transcripts (Iyyer et al 2014;Chen et al 2017;Lai et al 2020), and tweets from official sources like legislators, news agencies, and politicians (Xiao et al 2023), which typically adhere to strict grammatical rules, analyzing tweets from ordinary users presents unique challenges. These challenges include the brevity of the text, the absence of contextual information, and the frequent use of emoticons, abbreviations, and hashtags.…”

“…Given the sheer volume of data, traditional manual annotation meth-ods like crowdsourcing and editorial review (Li and Goldwasser 2019) become impractical. Some studies (Xiao et al 2023;Iyyer et al 2014;Chen et al 2017) have resorted to annotating texts based on the ideological positions of authors or speakers. However, for personal Twitter accounts, political ideologies are often undisclosed.…”