Scholarlink

The internet has become a crucial component of our lives in the contemporary digital age, placing a multitude of knowledge at the fingertips of the people. The rise of social media platforms and news websites has revolutionized how news is disseminated and enabled unequaled accessibility. However, this newly discovered convenience has also given rise to unscrupulous people wanting to benefit from misleading practices, such as clickbait headlines and the distribution of fake information. In our work solving the issue of recognizing false news, we developed a hybrid strategy, merging classic machine learning with sophisticated transfer learning models. Rigorous experimentation incorporated logistic regression, SVM, and neural networks like Bi-LSTM. Notably, logistic regression emerged as a top performer with an accuracy of 90.62%. These findings underline the efficacy of varied approaches in recognizing fraudulent material. We thoroughly tweaked our dataset to increase model performance. As contributors to the ongoing struggle against disinformation, our study underscores the potential outcomes gained through a diverse strategy. Looking ahead, our emphasis includes incorporating more complex models and researching real-time detection technologies to enhance defenses against the ever-evolving field of false news. This study indicates a step forward in employing a variety of approaches to solve the complicated and ubiquitous problem of recognizing and fighting disinformation in today’s digital age.

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

Enhancing Fake News Detection Through Machine Learning and Transfer Learning Methods

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Md Jakir Hossain,
  • Tashfia Choudhury,
  • Maleha Israt Chowdhury,
  • Sonjoy Chandra Mohonto,
  • Md Tanvir Chowdhury,
  • Habibur Rahman,
  • Monjurul Islam Sumon

摘要

The internet has become a crucial component of our lives in the contemporary digital age, placing a multitude of knowledge at the fingertips of the people. The rise of social media platforms and news websites has revolutionized how news is disseminated and enabled unequaled accessibility. However, this newly discovered convenience has also given rise to unscrupulous people wanting to benefit from misleading practices, such as clickbait headlines and the distribution of fake information. In our work solving the issue of recognizing false news, we developed a hybrid strategy, merging classic machine learning with sophisticated transfer learning models. Rigorous experimentation incorporated logistic regression, SVM, and neural networks like Bi-LSTM. Notably, logistic regression emerged as a top performer with an accuracy of 90.62%. These findings underline the efficacy of varied approaches in recognizing fraudulent material. We thoroughly tweaked our dataset to increase model performance. As contributors to the ongoing struggle against disinformation, our study underscores the potential outcomes gained through a diverse strategy. Looking ahead, our emphasis includes incorporating more complex models and researching real-time detection technologies to enhance defenses against the ever-evolving field of false news. This study indicates a step forward in employing a variety of approaches to solve the complicated and ubiquitous problem of recognizing and fighting disinformation in today’s digital age.