HullRyan / ITCS-5122-BERT-Sentimental-LIAR

ITCS 5122 - BERT Sentimental LIAR

Final Project/ Paper for Visual Analytics - ITCS 5122 University of North Carolina at Charlotte

The following work flow follows an implementation of a model found in Sentimental LIAR presented by Bibek Upadhayay and Vahid Behzadan

This follows the implementation by the authoes found here,
with minor changes to run locally without errors, and extra explantions in the code.

Documentation

View the iPython notebook here

The model included is an implementation of BERT-CNN, setup for 2-way classification on the Sentimental LIAR dataset.

Currently, it will run 2 training epochs, but there is further room for improvement here and in the hyperparameters.

Acknowledgements

• Sentimental LIAR: Extended Corpus and Deep Learning Models for Fake Claim Classification
• SentimentalLIAR