The purpose of the app is to train a model based on any two given Twitter users tweets. Upon receiving new, unseen text, the model will predict who was more likely to have written the text by comparing vectorized versions of each user's body of tweets.
-
Access the Twitter API
- Apply for elevated status to allow higher volume of tweets per month.
-
models.py: creates schema for SQLite database using SQLAlchemy -
predict.py: Looks at two given users' tweets, vectorizes them with Spacy -
twitter.py: Accesses Twitter API to add new users to the database from given user input. Checks to see if user is already in the database. Denotes current tweet ID for later reference when checking for updates. Also providesvectorize_tweet()function which applies Spacy vectorizations. -
app.py: Provides deployed Flask URL endpoints for viewing user tweets, resetting database, adding users to the database, updating existing users' tweets, and comparing users to generate a prediction for output. -
TODO: Update logistic regression model. Evaluate performance of other classificiation models on this task.
