Powered by OpenAIRE graph
Found an issue? Give us feedback
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/ Osuva (University of...arrow_drop_down
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
IEEE Access
Article . 2024 . Peer-reviewed
License: CC BY
Data sources: Crossref
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
IEEE Access
Article . 2024
Data sources: DOAJ
image/svg+xml art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos Open Access logo, converted into svg, designed by PLoS. This version with transparent background. http://commons.wikimedia.org/wiki/File:Open_Access_logo_PLoS_white.svg art designer at PLoS, modified by Wikipedia users Nina, Beao, JakobVoss, and AnonMoos http://www.plos.org/
Research.fi
Article . 2024 . Peer-reviewed
License: CC BY
Data sources: Research.fi
https://dx.doi.org/10.60692/ys...
Other literature type . 2024
Data sources: Datacite
https://dx.doi.org/10.60692/e4...
Other literature type . 2024
Data sources: Datacite
versions View all 5 versions
addClaim

This Research product is the result of merged Research products in OpenAIRE.

You have already added 0 works in your ORCID record related to the merged Research product.

Utilizing Ensemble Learning for Detecting Multi-Modal Fake News

استخدام التعلم الجماعي للكشف عن الأخبار المزيفة متعددة الوسائط
Authors: Muhammad Luqman; Muhammad Faheem; Waheed Yousuf Ramay; Malik Khizar Saeed; Majid Bashir Ahmad;

Utilizing Ensemble Learning for Detecting Multi-Modal Fake News

Abstract

La diffusion de fausses nouvelles est devenue un problème critique ces dernières années en raison de l'utilisation intensive des plateformes de médias sociaux. Les fausses histoires peuvent devenir virales rapidement, atteignant des millions de personnes avant qu'elles ne puissent être moquées, c'est-à-dire une fausse histoire affirmant qu'une célébrité est morte alors qu'elle est encore en vie. Par conséquent, la détection des fausses nouvelles est essentielle pour maintenir l'intégrité de l'information et contrôler la désinformation, la polarisation sociale et politique, l'éthique des médias et les menaces à la sécurité. Dans cette perspective, nous proposons une détection de fausses nouvelles multimodales basée sur l'apprentissage d'ensemble. Tout d'abord, il exploite un ensemble de données accessibles au public, Fakeddit, composé de plus d'un million d'échantillons de fausses nouvelles. Ensuite, il exploite les techniques de traitement du langage naturel (NLP) pour le prétraitement des informations textuelles des actualités. Ensuite, il mesure le sentiment à partir du texte de chaque nouvelle. Après cela, il génère des intégrations pour le texte et les images des actualités correspondantes en tirant parti des représentations de l'encodeur bidirectionnel visuel des transformateurs (V-BERT), respectivement. Enfin, il transmet les intégrations au modèle d'ensemble d'apprentissage profond pour la formation et les tests. La technique d'évaluation par 10 est utilisée pour vérifier la performance de l'approche proposée. Les résultats de l'évaluation sont significatifs et surpassent les approches de pointe avec l'amélioration des performances de 12,57 %, 9,70 %, 18,15 %, 12,58 %, 0,10 et 3,07 en précision, précision, rappel, score F1, coefficient de corrélation de Matthews (MCC) et Odds Ratio (OR), respectivement.

La difusión de noticias falsas se ha convertido en un problema crítico en los últimos años debido al uso extensivo de las plataformas de redes sociales. Las historias falsas pueden volverse virales rápidamente, llegando a millones de personas antes de que puedan ser burladas, es decir, una historia falsa que afirma que una celebridad ha muerto cuando todavía está viva. Por lo tanto, detectar noticias falsas es esencial para mantener la integridad de la información y controlar la desinformación, la polarización social y política, la ética de los medios y las amenazas a la seguridad. Desde esta perspectiva, proponemos una detección de noticias falsas multimodales basada en el aprendizaje conjunto. En primer lugar, explota un conjunto de datos disponibles públicamente, Fakeddit, que consta de más de 1 millón de muestras de noticias falsas. A continuación, aprovecha las técnicas de procesamiento del lenguaje natural (PNL) para preprocesar la información textual de las noticias. Luego, mide el sentimiento a partir del texto de cada noticia. Después de eso, genera incrustaciones para texto e imágenes de las noticias correspondientes mediante el aprovechamiento de las Representaciones de Codificadores Visuales Bidireccionales de Transformadores (V-BERT), respectivamente. Finalmente, pasa las incrustaciones al modelo de conjunto de aprendizaje profundo para capacitación y pruebas. La técnica de evaluación de 10 veces se utiliza para verificar el rendimiento del enfoque propuesto. Los resultados de la evaluación son significativos y superan los enfoques más avanzados, con una mejora del rendimiento del 12,57%, 9,70%, 18,15%, 12,58%, 0,10 y 3,07 en precisión, recordación, puntuación F1, coeficiente de correlación de Matthews (MCC) y razón de probabilidades (OR), respectivamente.

The spread of fake news has become a critical problem in recent years due extensive use of social media platforms. False stories can go viral quickly, reaching millions of people before they can be mocked, i.e., a false story claiming that a celebrity has died when he/she is still alive. Therefore, detecting fake news is essential for maintaining the integrity of information and controlling misinformation, social and political polarization, media ethics, and security threats. From this perspective, we propose an ensemble learning-based detection of multi-modal fake news. First, it exploits a publicly available dataset Fakeddit consisting of over 1 million samples of fake news. Next, it leverages Natural Language Processing (NLP) techniques for preprocessing textual information of news. Then, it gauges the sentiment from the text of each news. After that, it generates embeddings for text and images of the corresponding news by leveraging Visual Bidirectional Encoder Representations from Transformers (V-BERT), respectively. Finally, it passes the embeddings to the deep learning ensemble model for training and testing. The 10-fold evaluation technique is used to check the performance of the proposed approach. The evaluation results are significant and outperform the state-of-the-art approaches with the performance improvement of 12.57%, 9.70%, 18.15%, 12.58%, 0.10, and 3.07 in accuracy, precision, recall, F1-score, Matthews Correlation Coefficient (MCC), and Odds Ratio (OR), respectively.

أصبح انتشار الأخبار المزيفة مشكلة حرجة في السنوات الأخيرة بسبب الاستخدام المكثف لمنصات وسائل التواصل الاجتماعي. يمكن أن تنتشر القصص الكاذبة بسرعة، وتصل إلى ملايين الأشخاص قبل أن يتم السخرية منهم، أي قصة كاذبة تدعي أن أحد المشاهير قد مات عندما كان لا يزال على قيد الحياة. لذلك، يعد اكتشاف الأخبار المزيفة أمرًا ضروريًا للحفاظ على سلامة المعلومات والتحكم في المعلومات الخاطئة والاستقطاب الاجتماعي والسياسي وأخلاقيات وسائل الإعلام والتهديدات الأمنية. من هذا المنظور، نقترح اكتشافًا قائمًا على التعلم الجماعي للأخبار المزيفة متعددة الوسائط. أولاً، تستغل مجموعة بيانات متاحة للجمهور Fakeddit تتكون من أكثر من مليون عينة من الأخبار المزيفة. بعد ذلك، تستفيد من تقنيات معالجة اللغة الطبيعية (NLP) لمعالجة المعلومات النصية للأخبار مسبقًا. ثم يقيس المشاعر من نص كل خبر. بعد ذلك، يقوم بإنشاء تضمينات للنصوص والصور للأخبار المقابلة من خلال الاستفادة من تمثيلات التشفير المرئي ثنائي الاتجاه من المحولات (V - BERT)، على التوالي. أخيرًا، يمرر التضمينات إلى نموذج مجموعة التعلم العميق للتدريب والاختبار. تُستخدم تقنية التقييم المكونة من 10 أضعاف للتحقق من أداء النهج المقترح. نتائج التقييم كبيرة وتتفوق على أحدث الأساليب مع تحسين الأداء بنسبة 12.57 ٪ و 9.70 ٪ و 18.15 ٪ و 12.58 ٪ و 0.10 و 3.07 في الدقة والدقة والتذكر ودرجة F1 ومعامل ارتباط ماثيوز (MCC) ونسبة الاحتمالات (OR)، على التوالي.

Country
Finland
Keywords

FOS: Computer and information sciences, Exploit, Artificial intelligence, Sociology and Political Science, bagged CNN, convolutional neural network, Social Sciences, fi=Tietotekniikka|en=Computer Science|, Boosting (machine learning), Detection and Prevention of Phishing Attacks, Social media, Characterization and Detection of Android Malware, Computer security, Ensemble learning, Machine learning, boosted CNN, Information retrieval, Polymer chemistry, Preprocessor, Fake News, Natural language processing, multi-modal fake news, The Spread of Misinformation Online, Deep learning, Autoencoder, Rumor Detection, Computer science, 004, TK1-9971, Encoder, World Wide Web, Chemistry, Operating system, classification, Fake news, Computer Science, Physical Sciences, Signal Processing, Misinformation, Internet privacy, Electrical engineering. Electronics. Nuclear engineering, Modal, Information Systems

  • BIP!
    Impact byBIP!
    selected citations
    These citations are derived from selected sources.
    This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    13
    popularity
    This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
    Top 10%
    influence
    This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
    Average
    impulse
    This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
    Top 10%
Powered by OpenAIRE graph
Found an issue? Give us feedback
selected citations
These citations are derived from selected sources.
This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Citations provided by BIP!
popularity
This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network.
BIP!Popularity provided by BIP!
influence
This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically).
BIP!Influence provided by BIP!
impulse
This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network.
BIP!Impulse provided by BIP!
13
Top 10%
Average
Top 10%
Green
gold