you are viewing a single comment's thread.

view the rest of the comments →

[–]pancakewaffle 7 insightful - 1 fun7 insightful - 0 fun8 insightful - 1 fun -  (1 child)

This is both high level in description and low level in implementation so I won't dwell on it long but, crawl for author and user ids (probably in base), target those ids in intervals per api request limits, run them through a natural language set and train them. That way you can compare probabilities of a user against a suspected alt based on writing habits. It's also a bit overkill haha.

[–]jjdub7Gay Male Guest Commentator 2 insightful - 1 fun2 insightful - 0 fun3 insightful - 1 fun -  (0 children)

Oooh, interesting. Like a stylization attack, or whatever its called, as a user's fingerprint. Always on the hunt for new pet projects to expand on what I have, simply for skills development purposes - definitely adding a deep dive on this to my backlog.