Subscribe
Sign in
Botson — Detect bots in your twitter feed.
14
Replies
@sbmarkb Yes, this is a current limitation of the modeling, e.g. organizational accounts like "@barackobama" tend to score high (you can read the FAQ here: https://botometer.iuni.iu.edu/#!...). The reality is that it's difficult to classify bots. We're working on improving this. Thanks for your feedback!
@andyjiang thanks for that. I did read the FAQ. I’m not questioning your implementation, rather trying to understand. For instance, Mark Knoller & Dave Weigel are reporters; they write their tweets - they just happen to be writing about trending topics with high frequency. As a PM, I’m curious as to the decision not to implicitly trust a verified account as “not a bot” when classifying the content. I can’t think of a verified bot, but if they existed it would seem there would be significant false positives (negative value) in the detection of a few accounts. I tend to follow reporters, politicians, musicians... 2/3 of those being detected as bots renders the tool unusable for me.
@sbmarkb Hey no problem. This is great feedback, so thank you for sharing. Understand that it's unusable, hopefully we figure out the right way to improve the model :)
@andyjiang what a great idea you have going, though, and rest assured I have the extension still installed. Will definitely keep trying it out as you work out the kinks.
Hey, great idea for a product! How exactly do you determine what is a bot? Any specific technology?
Botson