Training a model to detect fake news articles.
These notebooks are part of a capstone project to identify text features that distinguish each of fake news articles, real news and satire from each other. My teammates worked on comparisons of real news to satire and satire to fake news. My part was to work on distinguishing fake news articles from real news.
Notebook A processes the raw data, transforming it into a format that can be used by the model. Then notebook B does most of the work, including identifying which text features were most relevant in distinguishing the fake news articles from the real ones.
The following two libraries are needed:
!pip install empath
!pip install scattertext
These libraries are recommended by the Scattertext authors, "in order to take full advantage of Scattertext", but might not need to be installed.
!pip install jieba
!pip install spacy
!pip install astropy
!pip install flashtext
!pip install gensim
!pip install umap-learn
Ali Tobah tobah@umich.edu
Free for personal or educational not for profit use. Otherwise, please contact the author. Please indicate the source in any case. Feel free to leave comments or suggestions.