A small API that uses a trained scikit-learn classifier to perform sentiment analysis on tweet-like objects. These tweets
are added to a MySQL database 'Tweets'.
A scikit-learn MultinomialNB classifier trained on the Sentiment140 tweets dataset is used to classify
tweets.
Technologies used:
Flaskscikit-learnmysql-connectorHTMLJavascriptpandas
To run:
- Change MySQL credentials as needed in
assets/config.py flask run(Note the port given, might be e.g. 8000)- That's it!
API Functions:
- Post a tweet
Tweets can be posted e.g.:
curl -i -H "Content-Type: application/json" -X POST -d '{"author":"@jake", "content":"why am I ugly"}' http://localhost:5000/api/tweet
Tweets must be in JSON format and have an author (e.g. @jake) and content (the tweet itself). If valid, the sentiment ("Positive"/"Negative") of the tweet is returned and the tweet is added to the 'Tweets' database.
- View a tweet
Tweets can be viewed by ID e.g. id=13:
curl -i http://localhost:5000/api/tweet/13
If the tweet is found, the tweet is returned else a 404 is returned
- View all tweets
curl -i http://localhost:5000/api/tweets/all
All tweets (all records of the Tweets database) are returned
- View all tweets by an author e.g.
author=@jake
curl -i http://localhost:5000/api/tweets/@jake
If tweet(s) by that author are found, the tweet(s) is/are returned else a 404 is returned
- View all tweets by sentiment e.g.
sentiment=positive
curl -i http://localhost:5000/api/tweets/sentiment/positive
If tweet(s) of that sentiment are found, the tweet(s) is/are returned else a 404 is returned
- Delete a tweet
Tweets can be deleted by ID e.g. id=12:
curl -X DELETE http://localhost:5000/api/tweet/12
If tweet is found, it is deleted from Tweets.
For fun, one can also navigate to http://localhost:5000/ and submit a tweet. The sentiment of the tweet is then displayed.
Note: Tweets submitted this way are not added to the Tweets database.