Fake News Detector AI
Documentation: https://www.stuvia.com/en-nl
https://www.studypool.com/
Source code : https://www.piecex.com/
then search by application name.
=========================================
In this video, we introduce our solution for detecting fake news using a supervised machine learning algorithm. Our model categorizes news articles as either fake or true based on various factors such as words, sentences, origins, and titles. We utilized a labeled dataset which was manually classified and registered to achieve a 50% accuracy rate in the initial model. For training, we used 80% of the prepared dataset to train the model and tested the remaining 20% to complete our model.
We also applied feature selection techniques to select the most suitable features for maximum accuracy based on the results of the Confusion Matrix. Our model utilizes the Naive Bayes classifier and Random Forest classifier to classify text-based information with high accuracy.
The software has functional requirements for both users and developers. Users can upload suspicious news articles and get predictions from the system if the information is false or true. They can also review the database to see the prediction history and print the outcomes. Developers, on the other hand, preprocess the labeled dataset, utilize NLP NLTK libraries and SAFAR v2 library to get rid of noise and extract features, split the dataset for training and testing, and create the classification model and confusion matrix.
Our software is operated on Windows 10, utilizing an environment named Anaconda 3 with Python language and Scikit-learn 0.21 library for ML pattern creation. We also used Jupyter version 4.3 for operating algorithms codes and dataset preprocessing. The minimum hardware requirement for this software is I7 core and 8GB RAM.