BirdSpotter is a python package which provides an influence and bot detection toolkit for twitter. You should use this package if you have tweet dumps and you are interested in:
- The probability that a user is actually a bot.
- How influential a user is compared to their peers.
BirdSpotter is designed to be simple to use for researchers and people who have a limited coding experience.
BirdSpotter is available though pip using pip install birdspotter within a terminal/cmd.
BirdSpotter's primary function is to take a whole tweet dump and determine botness and influence metrics for each user. We expose this functionality through a class as follows.
from BirdSpotter import BirdSpotter
bs = BirdSpotter()We can extract the information from our dumps with extractTweets()
bs.extractTweets('/path/to/tweet/dump.json')To determine the 'botness' of the users in the dump we need to provide some bot training examples. If you don't have bot examples train.pickle is provided in the repository and can be downloaded. Alternatively, you can label your own data with the help of getBotAnnotationTemplate()
bs.trainClassifierModel("train.pickle")Finally, to get the individual users metrics we can run methods like getBotness(), which will return a pandas dataframe of each user with their 'botness' metric.
bs.getBotness()For more specific documentation refer to the tutorial or the documentation.
The bot detection system used in this project is based on the work of Yifei Zhang's (The Bot Among Us: Disrupting Truth and Reason Through Online Social media, 2018).
Both dataset and code are distributed under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license, a copy of which can be obtained following this link. If you require a different license, please contact us at Marian-Andrei@rizoiu.eu