/fake-news-challenge-1

Our Project for the Unstructured Information Processing course on Fake News Challenge (http://www.fakenewschallenge.org/)

Primary LanguageJupyter Notebook

fake-news-challenge-1

Our Project for the Unstructured Information Processing course on Fake News Challenge (http://www.fakenewschallenge.org/)

A machine learning approach to stance detection. The problem statement can be found at the official website introducing the challenge. Explanations for our wok can be found in Report.pdf. This was our end-semester project for Unstructured Information Processing (CS-406) at Ashoka University (in Monsoon 2019).

Authors: Prakhar Jain, Dhruv Agarwal

How to use

  • Use main.ipynb to go through the code. This file is well commented and hence self-explanatory. Run only the cells inside a "demo" heading to see predictions.
  • model_ru.h5 and vectorizer_ru.pickle are the model and vectorizer for the first sub-task (related-unrelated).
  • model_add.h5 and vectorizer_add.pickle are the model and vectorizer for the second sub-task (agree-disagree-discuss).