Fake News Classifier
Abstract
In this digital era of smartphones and the internet, the fake news spreads like wildfire, and causes more damage to the society. That is where we think that, the need for fake news classification arises due to the spread of misinformation and false information through various media platforms, which can have serious consequences such as impacting public opinion and decision-making, causing harm to individuals or groups, and eroding trust in information sources.
Problem Statement:
So for our project we would like work on fake news binary classification problem where we will be classifying whether the news is fake or real using the title and news text which is available in our dataset that we got from Kaggle (Dataset: https://www.kaggle.com/c/fake-news/data#).
Why we chose this dataset?
As this dataset contains enough rows (more than 18000+) news articles for classification, and this is also an open source dataset from a Kaggle Competition.
Proposed Solution:
The solution will include an optimized classifier that will return a positive/negative output to classify a given news article as real/fake respectively. Here we will be developing a baseline RNN model with GRU and LSTM along with hyperparameter tuning to find the most accurate model and study how various models and hyperparameters will impact the classification. Here we are planning to use different activation functions with LSTM and GRU (this is just an example). We are also planning to apply different deep learning approach that has been taught in class, and we will be coming out with the best model and hyperparameters for classification. And we will be comparing our results as well.