Fake News Analysis

A fake news classifier built with sckit-learn and a bit of NLTK. For a brief overview, check out the project website or our Jupyter notebook.

Dependencies

python==2.7.14
Flask==1.0.2
gunicorn
numpy==1.10.4
pandas==0.17.1
scikit-learn==0.17
pyldavis==2.1.2
nltk==3.3
scipy==0.13.3
gensim

To run the Flask app locally, run flask run.

LDA Visualizations

A quick note on generating the LDA visualizations locally. You may have difficulty running LdaMulticore(corpus,num_topics=10,id2word=article_dict,workers=2) in Jupyter. As an alternative, you can run the following commands in your terminal to generate the LDA models for the general corpus, fake articles, and real articles respectively:

python tinker.py
python fake_tinker.py
python real_tinker.py