A COMPARATIVE STUDY OF MAJORITY AND WEIGHTED VOTING ENSEMBLE TECHNIQUES FOR IMPROVING THE ACCURACY OF FALSE INFORMATION DETECTION

Abstract

A method of information dissemination that had never been seen before the digital era has emerged as a result of the development of the World Wide Web and the use of social media platforms like Facebook, Twitter & Instagram. Through these social media sites, a significant quantity of information may be fraudulent. Therefore, it is necessary to monitor this data. By feeding a new article or fact to the model, we may use a technique called machine learning-based fake news identification to determine its veracity. We will preprocess the data (in this example, text) before training the dataset. Preprocessing mostly entails eliminating redundant data. The dataset is then divided into two sections for testing and training. The data will next be vectorized using a variety of vectorization methods, including Countvectorizer, TF-IDF vectorizer, and n-grams. The various classifiers (such as random forest, decision tree, logistic regression etc.) are then trained using the vectorized data. The accuracy of cutting-edge false news detection is then increased by incorporating these results into ensemble models. The timeline of a dataset has an impact on how accurate the model is since newer information is not included in older datasets, making it impossible for the model to effectively forecast the veracity of newer information. After training, the model may be quickly tested using the testing dataset, and it will then be ready for usage.