Scholarlink

<p>Fake news spreads rapidly across digital platforms and affects public trust and informed decision-making in many regions. This work presents the Advancing Regional Language Fake News Detection through Interpretable Generalized Additive Neural Networks combined with FastText for Efficient Classification (AFND- IGANN- EC) model for detecting fake news in Tamil and Hindi. The method uses natural language processing (NLP)-based preprocessing to clean the text and extract essential features. Multi-Agent Cubature Kalman Optimizer (MACKO) selects the most relevant textual contextual and semantic information. Interpretable Generalized Additive Neural Network (IGANN) combined with FastText performs the final classification of real and fake news. Gooseneck Barnacle Optimization Algorithm (GBOA) tunes the model to achieve higher accuracy and lower error. Experimental evaluation shows that the proposed approach offers better accuracy higher precision and reduced transition latency when compared to existing fake news detection methods. The results confirm that AFND–IGANN–EC provides an effective interpretable and robust solution for regional language fake news detection.&#xa0;The proposed method attains 6.75%, 7.94%, and 8.22% higher accuracy compared to others methods such as Secure ensemble-based algorithm for fake news detection utilizing blockchain (SEB-FND-BC), multimodal fake news detection depending upon multi-granularity feature fusion and contrastive learning (MFND-MGFF-CL) and Linguistic feature fusion for detection of Arabic fake news and named entity recognition utilizing reinforcement learning and swarm optimization (LFF-AFND-RLSO) respectively.</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

Advancing regional language fake news detection through interpretable generalized additive neural networks combined with fast text for efficient classification

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • V. Rathinapriya,
  • M. Bhanumathi,
  • J. Kalaivani

摘要

Fake news spreads rapidly across digital platforms and affects public trust and informed decision-making in many regions. This work presents the Advancing Regional Language Fake News Detection through Interpretable Generalized Additive Neural Networks combined with FastText for Efficient Classification (AFND- IGANN- EC) model for detecting fake news in Tamil and Hindi. The method uses natural language processing (NLP)-based preprocessing to clean the text and extract essential features. Multi-Agent Cubature Kalman Optimizer (MACKO) selects the most relevant textual contextual and semantic information. Interpretable Generalized Additive Neural Network (IGANN) combined with FastText performs the final classification of real and fake news. Gooseneck Barnacle Optimization Algorithm (GBOA) tunes the model to achieve higher accuracy and lower error. Experimental evaluation shows that the proposed approach offers better accuracy higher precision and reduced transition latency when compared to existing fake news detection methods. The results confirm that AFND–IGANN–EC provides an effective interpretable and robust solution for regional language fake news detection. The proposed method attains 6.75%, 7.94%, and 8.22% higher accuracy compared to others methods such as Secure ensemble-based algorithm for fake news detection utilizing blockchain (SEB-FND-BC), multimodal fake news detection depending upon multi-granularity feature fusion and contrastive learning (MFND-MGFF-CL) and Linguistic feature fusion for detection of Arabic fake news and named entity recognition utilizing reinforcement learning and swarm optimization (LFF-AFND-RLSO) respectively.