twitter-sentiment is a Python library leveraging NLP algorithm and the Twitter API to classify the sentiment of a tweet.
Installing twitter-sentiment is simple, you just have to use pip. ::
pip install twitter-sentiment
Documentation is available at twitter-sentiment.readthedocs.io
twitter-sentiment lets you classify a tweet/list of tweets as positive (1) or negative (0). twitter-sentiment then calculates and returns the ration of positive tweets. To classify a tweet, twitter-sentiment leverages TextBlob Naive Bayes NLP library. More information can be found at textblob.readthedocs.io
twitter-sentiment uses circleci as a continuous integration tool. Pushing a new git tag to the remote repository will trigger circleci workflow and:
- validate the test in /test/test_twitterSentiment.py
- check for a match between the
VERSION
variable in the setup.py file and the git tag version.
If all tests pass, the build will be automatically uploaded to the pypi server.