The adaptive community-response (ACR) method for collecting misinformation on social media

Journal Of Big Data, Год журнала: 2024, Номер 11(1)

Опубликована: Фев. 24, 2024

Abstract Social media can be a major accelerator of the spread misinformation, thereby potentially compromising both individual well-being and social cohesion. Despite significant recent advances, study online misinformation is relatively young field facing several (methodological) challenges. In this regard, detection has proven difficult, as large-scale data streams require (semi-)automated, highly specific therefore sophisticated methods to separate posts containing from irrelevant posts. present paper, we introduce adaptive community-response (ACR) method, an unsupervised technique for collection on Twitter (now known ’X’). The ACR method based previous findings showing that users occasionally reply with fact-checking by referring sites (crowdsourced fact-checking). first step, captured such misinforming but fact-checked tweets. These tweets were used in second step extract linguistic features (keywords), enabling us collect also those not at all third step. We initially mathematical framework our followed explicit algorithmic implementation. then evaluate basis comprehensive dataset consisting $$>25$$ > 25 million tweets, belonging $$>300$$ 300 stories. Our evaluation shows useful extension pool field, researchers more comprehensively. Text similarity measures clearly indicated correspondence between claims false stories even though performance was heterogeneously distributed across A baseline comparison showed detect story-related comparable degree, while being sensitive different types tweets: Fact-checked tend driven high outreach (as number retweets), whereas sensitivity extends exhibiting lower outreach. Taken together, ACR’s capacity valuable methodological contribution (i) adoption prior, pioneering research (ii) well-formalized (iii) empirical foundation via set indicators.

Язык: Английский

---

Online disinformation in the 2020 U.S. election: swing vs. safe states

EPJ Data Science, Год журнала: 2024, Номер 13(1)

Опубликована: Март 26, 2024

Язык: Английский

---

Patterns of partisan toxicity and engagement reveal the common structure of online political communication across countries

Nature Communications, Год журнала: 2024, Номер 15(1)

Опубликована: Ноя. 14, 2024

Existing studies of political polarization are often limited to a single country and one form polarization, hindering comprehensive understanding the phenomenon. Here we investigate patterns online across nine countries (Canada, France, Germany, Italy, Poland, Spain, Turkey, UK, USA), focusing on structure interaction networks, use toxic language targeting out-groups, how these factors relate user engagement. First, show that networks structurally polarized Twitter (currently X). Second, reveal out-group interactions, defined by network, more than in-group indicative affective polarization. Third, interactions receive lower engagement interactions. Finally, identify common ally-enemy in mentions apolitical mentions, highlight between politically engaged accounts rarely reciprocated. These results hold represent step towards stronger cross-country Identifying is important for its root cause. Here, using data from 9 countries, authors mentions.

Язык: Английский

---

Nudging recommendation algorithms increases news consumption and diversity on YouTube

PNAS Nexus, Год журнала: 2024, Номер 3(12)

Опубликована: Ноя. 19, 2024

Abstract Recommendation algorithms profoundly shape users’ attention and information consumption on social media platforms. This study introduces a computational intervention aimed at mitigating two key biases in by influencing the recommendation process. We tackle interest bias, or creating narrow nonnews entertainment diets, ideological directing more strongly partisan users to like-minded content. Employing sock-puppet experiment (n=8,600 sock puppets) alongside month-long randomized involving 2,142 frequent YouTube users, we investigate if nudging algorithm playing videos from verified ideologically balanced news channels background increases recommendations of news. additionally test providing input promotes diverse cross-cutting consumption. find that significantly sustainably both also minimizes consumption, particularly among conservative users. In fact, have stronger effects exposure than has subsequent recommendations. contrast, no observable Increased range survey outcomes (i.e. political participation, belief accuracy, perceived affective polarization, support for democratic norms), adding growing evidence limited attitudinal on-platform exposure. The does not adversely affect user engagement YouTube, showcasing its potential real-world implementation. These findings underscore influence wielded platform recommender

Язык: Английский

---

Metaverse maelstrom: Dissecting information dynamics and polarisation

Journal of Information Science, Год журнала: 2025, Номер unknown

Опубликована: Янв. 3, 2025

The Metaverse represents a collaborative virtual realm blending physical and digital realities, fostering limitless avenues for online interaction, discovery innovation. As technological strides propel immersive worlds to the forefront of social media platforms, scholarly interest in surges, prompting extensive discourse. Drawing from identity theory, this article introduces novel framework analysing polarisation within discussions on Metaverse, specifically X (Twitter). Leveraging multifaceted approach that integrates clustering, network analysis, text mining, our study delves into both group opinion dynamics surrounding Metaverse. Our findings uncover distinct community divisions structures, shedding light prevalent themes, such as ‘Non-Fungible Token (NFTs)’, ‘Virtual Products Collections’, ‘Blockchain Technology’, ‘Gaming’, ‘Financial Markets’ resonate public

Язык: Английский

---