NLP Classification for Satire detection

The genre of satire often falls through the cracks of real-fake news detection algorithms. This project aims to use Natural Language Processing (NLP) features as input data for various Machine Learning (ML) algorithms, and find a locally optimal combination of features for separating news headlines into real, fake and satirical news.

File Index

1. Code
   1. CNN_models.ipnyb - Python notebook for Convolution Neural Network (CNN) training
   2. Non_neural_models.ipnyb - Python notebook for non-neural model training (Support Vector Machine, Logistic regression)
   3. POS_tagger_analysis.ipnyb - Python notebook for Part-Of-Speech (POS) analysis on dataset
   4. README.md - this file
   5. RNN_models.ipnyb - Python notebook for Recurrent Neural Network (RNN) training
2. Writeup.pdf - Poster of the group project, detailing method, findings and analysis.

N.B. These notebook files were run in Google Colab, so the filepaths have to be edited according to the environment.