You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
milad532 ea31eccef4 Upload files 1 month ago
Deepseek Upload files 1 month ago
GPT-4o Upload files 1 month ago
LLaMa Upload files 1 month ago
Prompt_0 Upload files 1 month ago
Qwen Upload files 1 month ago
README.md Upload files 1 month ago
log_processor.py Upload files 1 month ago
news_dataset.csv Upload files 1 month ago
news_leaning_dataset.csv Upload files 1 month ago

README.md

Fake News Detection Using LLMs

πŸ“Œ Description

This repository contains the implementation of the paper β€œToward Fair and Effective Fake News Detection: Assessing Large Language Models.” The project focuses on evaluating the fairness and efficiency of Large Language Models (LLMs) in detecting fake news using a dataset of news articles classified by political leaning.

πŸš€ Features

  • Analysis of LLM biases in news classification
  • Evaluation of model fairness and accuracy
  • Benchmarking multiple LLMs (GPT-4o, LLaMa, Qwen, Deepseek)
  • News leaning classification (Democrat, Republican, Neutral, Varies)
  • Fake news detection using a labeled dataset

πŸ“– Usage

use the news_dataset.csv and news_leaning_dataset.csv as dataset and log_processor.py as processor for the outputs of each LLM.

πŸ› οΈ Technologies Used

  • Python
  • open-ai
  • Scikit-learn
  • Pandas & NumPy

πŸ“Š Dataset

The dataset consists of news articles labeled with political leanings and fact-checking results. The files include:

  • news_dataset.csv: Contains raw news articles with metadata with labeled POV.
  • news_leaning_dataset.csv: Labels news articles as Democrat, Republican, Neutral, or Varies with Labeled leanings.