Journal of Medical Internet Research, Год журнала: 2023, Номер 25, С. e49416 - e49416
Опубликована: Сен. 22, 2023
While there has been substantial analysis of social media content deemed to spread misinformation about electronic nicotine delivery systems use, the strategic use accusations undermine opposing views received limited attention.
Язык: Английский
Science, Год журнала: 2024, Номер 384(6699)
Опубликована: Май 30, 2024
Low uptake of the COVID-19 vaccine in US has been widely attributed to social media misinformation. To evaluate this claim, we introduce a framework combining lab experiments (total N = 18,725), crowdsourcing, and machine learning estimate causal effect 13,206 vaccine-related URLs on vaccination intentions Facebook users ( ≈ 233 million). We that impact unflagged content nonetheless encouraged skepticism was 46-fold greater than misinformation flagged by fact-checkers. Although reduced predicted significantly more when viewed, users’ exposure limited. In contrast, stories highlighting rare deaths after were among Facebook’s most-viewed stories. Our work emphasizes need scrutinize factually accurate but potentially misleading addition outright falsehoods.
Язык: Английский
Процитировано
30
Nature Medicine, Год журнала: 2024, Номер 30(6), С. 1559 - 1563
Опубликована: Апрель 29, 2024
Abstract It is unclear how great a challenge pandemic and vaccine fatigue present to public health. We assessed perspectives on coronavirus disease 2019 (COVID-19) routine immunization as well trust in information sources future preparedness survey of 23,000 adults 23 countries October 2023. The participants reported lower intent get COVID-19 booster 2023 (71.6%), compared with 2022 (87.9%). A total 60.8% expressed being more willing vaccinated for diseases other than result their experience during the pandemic, while 23.1% less willing. Trust 11 selected each averaged 7 10-point scale one’s own doctor or nurse World Health Organization, averaging 6.9 6.5, respectively. Our findings emphasize that hesitancy challenges remain health practitioners, underscoring need targeted, culturally sensitive communication strategies.
Язык: Английский
Процитировано
27
Natural Language Processing Journal, Год журнала: 2024, Номер 6, С. 100053 - 100053
Опубликована: Янв. 5, 2024
Spreading misinformation and fake news about COVID-19 has become a critical concern. It contributes to lack of trust in public health authorities, hinders actions from controlling the virus's spread, risks people's lives. This study aims gain insights into types spread develop an in-depth analytical approach for analyzing news. combines idea Sentiment Analysis (SA) Topic Modeling (TM) improve accuracy topic extraction large volume unstructured texts by considering sentiment words. A dataset containing 10,254 headlines various sources was collected prepared, rule-based SA applied label with three tags. Among TM models evaluated, Latent Dirichlet Allocation (LDA) demonstrated highest coherence score 0.66 20 coherent negative sentiment-based topics 0.573 18 positive topics, outperforming Non-negative Matrix Factorization (NMF) (coherence: 0.43) Semantic (LSA) 0.40). The extracted experiments highlight that primarily revolves around COVID vaccine, crime, quarantine, medicine, political social aspects. research offers insight effects news, provides valuable method detecting misinformation, emphasizes importance understanding patterns themes protecting promoting scientific accuracy. Moreover, it can aid developing real-time monitoring systems combat extending beyond COVID-19-related enhancing applicability findings.
Язык: Английский
Процитировано
12
PLoS ONE, Год журнала: 2024, Номер 19(5), С. e0302201 - e0302201
Опубликована: Май 22, 2024
The world’s digital information ecosystem continues to struggle with the spread of misinformation. Prior work has suggested that users who consistently disseminate a disproportionate amount low-credibility content—so-called superspreaders—are at center this problem. We quantitatively confirm hypothesis and introduce simple metrics predict top superspreaders several months into future. then conduct qualitative review characterize most prolific analyze their sharing behaviors. Superspreaders include pundits large followings, media outlets, personal accounts affiliated those range influencers. They are primarily political in nature use more toxic language than typical user also find concerning evidence suggests Twitter may be overlooking prominent superspreaders. hope will further public understanding bad actors promote steps mitigate negative impacts on healthy discourse.
Язык: Английский
Процитировано
8
Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 422 - 437
Опубликована: Янв. 1, 2025
Язык: Английский
Процитировано
1
Erciyes İletişim Dergisi, Год журнала: 2025, Номер 12(1), С. 159 - 186
Опубликована: Янв. 30, 2025
Dezenformasyon, kitle iletişimindeki en önemli sorunlardan biri olma özelliğini sürdürmektedir. Bu sorun özellikle olağanüstü dönemlerde ciddi seviyelere ulaşıp tehlikeli bir hal almaktadır. Türkiye’de 11 ilde yıkıcı etki yapan Kahramanmaraş depreminde de dezenformasyonun büyük zararları ortaya çıkmıştır. Depremin hemen ardından yoğun dezenformasyon başladığı için, Dezenformasyonla Mücadele Merkezi ilk ay bültenlerinin yüzde 93’ünü deprem konusundaki hatalı içerikleri düzeltmeye ayırmıştır. çalışma da dezenformasyonla mücadele etmek için yapılan girişimlerin ne kadar başarılı olduğu, sorunsalından hareketle gerçekleştirilmiştir. Çalışmanın amacı çalışmalarının verimliliğini koyarak alınabilecek yeni tedbirler perspektif oluşturmaktır. Çalışmada, Dezenformasyon Bültenleri doküman analizi ile incelenmiş; X platformundaki gönderiler açık içerik değerlendirilmiş; içeriği ve motivasyonu ise mesaj çözümlemesi yöntemiyle analiz edilmiştir. Çalışma sosyal medyanın dezenformasyondaki ısrarını istatiksel anlamda koyması, dezenformasyonunun motivasyonunu geniş alana yayılmasını göstermesi bakımından önemlidir. Çalışmada örneklem olarak seçilen 39 dezenformasyonun, 45 milyon 808 bin görüntülendiği, içeriklerin 87 oranında düzeltilmediği ya silinmediği çoğunlukla siyasi motivasyonlarla üretildiği görülmüştür. Sosyal yanlış bilgiyi düzeltmeme sorunu kez daha gözler önüne serilmiştir. neticesinde çarpıtılmış bilgilerin düzeltilmesi farklı stratejilere ihtiyaç duyulduğu anlaşılmıştır. konudaki mevcut öneriler sıralandıktan sonra sorunun çözümü adımlarla ilgili tartışma yapılmıştır.
Процитировано
0
Proceedings of the International AAAI Conference on Web and Social Media, Год журнала: 2023, Номер 17, С. 890 - 901
Опубликована: Июнь 2, 2023
Social media provide a fertile ground where conspiracy theories and radical ideas can flourish, reach broad audiences, sometimes lead to hate or violence beyond the online world itself. QAnon represents notable example of political that started out on social but turned mainstream, in part due public endorsement by influential figures. Nowadays, conspiracies often appear news, are rhetoric, espoused significant swaths people United States. It is therefore crucial understand how such took root online, what led so many users adopt its ideas. In this work, we propose framework exploits both interaction content signals uncover evidence user radicalization support for QAnon. Leveraging large dataset 240M tweets collected run-up 2020 US Presidential election, define validate multivariate metric radicalization. We use separate distinct, naturally-emerging, classes behaviors associated with processes, from self-declared supporters hyper-active promoters. also analyze impact Twitter's moderation policies interactions among different classes: discover aspects succeed, yielding substantial reduction received hyperactive accounts. But fails, showing amplifiers not deterred affected Twitter intervention. Our findings refine our understanding reveal effective ineffective moderation, call need further investigate role play spread conspiracies.
Язык: Английский
Процитировано
12
Опубликована: Сен. 9, 2023
Low uptake of the COVID-19 vaccine in US has been widely attributed to social media misinformation. To evaluate this claim, we introduce a framework combining lab experiments (total N=18,725), crowdsourcing, and machine learning estimate causal effect 13,206 vaccine-related URLs on vaccination intentions Facebook users (N≈233 million). We impact misinformation flagged by fact-checkers was 46X less than that unflagged content nonetheless encouraged skepticism. Although reduced significantly more when viewed, content’s exposure limited. In contrast, stories highlighting rare deaths following were among Facebook’s most-viewed stories. Our work emphasizes need scrutinize factually accurate but potentially misleading addition outright falsehoods.
Язык: Английский
Процитировано
11
EPJ Data Science, Год журнала: 2023, Номер 12(1)
Опубликована: Окт. 9, 2023
Abstract The detection of state-sponsored trolls operating in influence campaigns on social media is a critical and unsolved challenge for the research community, which has significant implications beyond online realm. To address this challenge, we propose new AI-based solution that identifies troll accounts solely through behavioral cues associated with their sequences sharing activity, encompassing both actions feedback they receive from others. Our approach does not incorporate any textual content shared consists two steps: First, leverage an LSTM-based classifier to determine whether account belong or organic, legitimate user. Second, employ classified calculate metric named “Troll Score”, quantifying degree exhibits troll-like behavior. assess effectiveness our method, examine its performance context 2016 Russian interference campaign during U.S. Presidential election. experiments yield compelling results, demonstrating can identify AUC close 99% accurately differentiate between organic users 91%. Notably, behavioral-based holds advantage ever-evolving landscape, where linguistic properties be easily mimicked by Large Language Models (LLMs): In contrast existing language-based techniques, it relies more challenging-to-replicate cues, ensuring greater resilience identifying campaigns, especially given potential increase usage LLMs generating inauthentic content. Finally, assessed generalizability various entities driving different information operations found promising results will guide future research.
Язык: Английский
Процитировано
11
Akademik Yaklaşımlar Dergisi, Год журнала: 2024, Номер 15(1 -Deprem Özel Sayısı-), С. 411 - 429
Опубликована: Янв. 20, 2024
Doğal afetler insan müdahalesi olmadan, beklenmeyen bir zamanda gerçekleşen ve yıkıcı sonuçlara sahip olabilen doğa olaylarıdır. Afetlerin doğal olarak kaotik süreci vardır bu nedenden yönetilmesi oldukça güçtür. Afetzedeler ile doğru iletişim hızlı karar verme afet sonucundaki olumsuzlukları azaltabilir. Günümüzde güçlü aracı sıklıkla kullanılan sosyal medya, yönetiminde kullanımı son derece önemlidir. Ancak medya belirli kontrol mekanizması anonim ortamlardır. Yazılan her paylaşım olmayabilir hatta art niyetli olabilmektedir. Bu çalışmada sonrası oluşturulan depremle ilgili yapan hesapların yaptığı paylaşımlar üzerinden analiz gerçekleştirilmiştir. Sosyal medyanın kullanımının en büyük engellerinden birisi olan hesap güvenirliğinin üzerine değerlendirme yapılmıştır. 6 Şubat 2023 Büyük Kahramanmaraş depreminden sonra ilk 7 günde 3.146 hesabın oluşturulduğu 6.724 tane görülmüştür. Bugün yapılan kontrollerde 5 üzeri %48’nin platform tarafından askıya alındığı veya kapatıldığı Hesapların mevcut durumda açık olanlarının ortalama 14 takipçi kazandığı Ayrıca tüm sırasıyla “Tepki/Dilek”, “Yardım Talebi” ve” Kurtarma kategorilerinde yaptıkları
Процитировано
4