Scholarlink

In a digital media-driven world, misinformation proliferates across online platforms. This study focuses on distinguishing real from fake news using advanced NLP techniques. We apply various word embeddings—TF-IDF, BERT, ALBERT, DistilBERT, GloVe, and RoBERTa—alongside machine learning algorithms like KNN, Random Forest, Logistic Regression, and deep learning models such as LSTM, GRU, BiLSTM, and a stacked LSTM-GRU model. We evaluate three scenarios: using only the title, only the content, and both for classification. Our results show that title-only classification reached 92% accuracy with the stacked model, content-only achieved 94%, and combining both improved accuracy to 96%. This highlights the effectiveness of hybrid deep learning techniques and multiple data sources in detecting fake news, emphasizing the need for sophisticated models to combat misinformation in today’s media landscape.

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

Precision in Fake News Detection: A Multi-embedding and Stacked Neural Network Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Shivam Bhadula,
  • Devashish Kartik,
  • Deepa Gupta

摘要

In a digital media-driven world, misinformation proliferates across online platforms. This study focuses on distinguishing real from fake news using advanced NLP techniques. We apply various word embeddings—TF-IDF, BERT, ALBERT, DistilBERT, GloVe, and RoBERTa—alongside machine learning algorithms like KNN, Random Forest, Logistic Regression, and deep learning models such as LSTM, GRU, BiLSTM, and a stacked LSTM-GRU model. We evaluate three scenarios: using only the title, only the content, and both for classification. Our results show that title-only classification reached 92% accuracy with the stacked model, content-only achieved 94%, and combining both improved accuracy to 96%. This highlights the effectiveness of hybrid deep learning techniques and multiple data sources in detecting fake news, emphasizing the need for sophisticated models to combat misinformation in today’s media landscape.