Scholarlink

<p>In emergencies, online media are among the main information sources; however, fake news spreads rapidly and erodes trust. Traditional methods for fake news detection are mostly based on text features and do not include mechanisms for contextual validation, making them vulnerable to adversarial attacks. Therefore, this work introduces a semantic-contextual approach to news verification that leverages deep learning techniques and real-world verification using Internet of Things (IoT) data. The model incorporates sequential learning and convolutional feature extraction to model text dependencies and important semantic features, respectively. To boost contextual trust evaluation, the model uses real-world evidence, such as sensor data streams, surveillance signals and geospatial metadata. This multimodal integration framework allows cross-referencing textual information with actual observations, thus mitigating semantic uncertainty and enhancing trust assessment. Extensive experiments on standard datasets show improved classification results, with an F1 score of 97%, accuracy of 97%, precision of 98.4% and recall of 98.25%. benchmarking shows steady improvement over traditional deep learning and probabilistic approaches. Also, the inclusion of real-world contextual cues greatly reduces classification error and improves verification in real time. This study shows that integrating semantic intelligence with evidence yields a robust and scalable approach to misleading information detection in the digital age.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

News authenticity verification and false information detection using deep learning and internet of things-based real-world validation

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月9日

查看全文
  • Yubei Qian

摘要

In emergencies, online media are among the main information sources; however, fake news spreads rapidly and erodes trust. Traditional methods for fake news detection are mostly based on text features and do not include mechanisms for contextual validation, making them vulnerable to adversarial attacks. Therefore, this work introduces a semantic-contextual approach to news verification that leverages deep learning techniques and real-world verification using Internet of Things (IoT) data. The model incorporates sequential learning and convolutional feature extraction to model text dependencies and important semantic features, respectively. To boost contextual trust evaluation, the model uses real-world evidence, such as sensor data streams, surveillance signals and geospatial metadata. This multimodal integration framework allows cross-referencing textual information with actual observations, thus mitigating semantic uncertainty and enhancing trust assessment. Extensive experiments on standard datasets show improved classification results, with an F1 score of 97%, accuracy of 97%, precision of 98.4% and recall of 98.25%. benchmarking shows steady improvement over traditional deep learning and probabilistic approaches. Also, the inclusion of real-world contextual cues greatly reduces classification error and improves verification in real time. This study shows that integrating semantic intelligence with evidence yields a robust and scalable approach to misleading information detection in the digital age.