smridhvarma/fake-news-detection

This repository houses a Streamlit web application for fake news detection. The app allows users to input a news article and predicts whether it is likely fake or real based on its content. It provides options to select different vectorizers (TF-IDF or Bag of Words) and classifiers (Linear SVM or Naive Bayes) to customize the prediction model.

Fake News Detection Webapp

Description

This Streamlit app is designed to detect whether a news article is likely fake or real based on its content. It allows users to input a news article, select a vectorizer and classifier, and then predicts the authenticity of the article.

Modules

Module 1: Import necessary packages

• streamlit: For creating the web application.
• numpy: For numerical computations.
• pandas: For data manipulation and analysis.
• sklearn: For machine learning functionalities.
• warnings: For ignoring warnings.
• streamlit_lottie: For displaying Lottie animations.

Module 2: Load the dataset

• Loads the dataset containing fake and real news articles from a CSV file.
• Converts the labels to binary format (0 for real, 1 for fake).

Module 3: Select Vectorizer and Classifier

• Allows users to select a vectorizer (TF-IDF or Bag of Words) and a classifier (Linear SVM or Naive Bayes) via the sidebar.

Module 4: Train the model

• Trains the selected classifier model using the chosen vectorizer and the loaded dataset.
• Caches the trained model for faster access.

Module 5: Streamlit app

• Sets page configuration including title, icon, and layout.
• Displays the title and a Lottie animation.
• Hides the Streamlit style for a cleaner interface.
• Provides a text area for users to input news articles.
• Upon clicking the "Check" button, predicts the authenticity of the input news article using the trained model and displays the result.

Usage

• Run the Streamlit app using the command: streamlit run main.py --client.showErrorDetails=false to remove cache error messages on the Streamlit interface.
• Input a news article into the text area.
• Select a vectorizer and classifier from the sidebar.
• Click the "Check" button to see the prediction result.