osome-iu / botometer-python

A Python API for Botometer by OSoMe
https://botometer.osome.iu.edu
MIT License
371 stars 59 forks source link

Threshold for determining Bot #35

Closed sjgiorgi closed 4 years ago

sjgiorgi commented 4 years ago

Is there a standard threshold for determining whether an account is a bot, assuming the returned scores are probabilities?

Which score is commonly used?

clayadavis commented 4 years ago

The cap scores are probabilities, the "raw" scores aren't. As far as a standard threshold, each individual needs to choose a threshold that works for them depending on their tolerance for false positive/negative results.

clayadavis commented 4 years ago

Use the CAP scores. The raw scores are there mostly for backwards-compatibility.

sjgiorgi commented 4 years ago

What is the different between english and universal? Is one preferred over the other?

clayadavis commented 4 years ago

english uses all of the features, while universal omits the content and sentiment features. Use english if you're sure your content is English-language, universal otherwise.