Idea: Automated filtering

[evanp]

Automated filtering with machine learning models are an important part of trust and safety on social networks.

Techniques like naive bayesian filtering have been used for detecting spam and other unwanted content in email. Akismet is a blog spam filtering system. ActivitySpam is a server for spam detection in Activity Streams 1.0 JSON.

Investigating how automated filtering fits into the ActivityPub ecosystem is a good use case for this working group.

[ThisIsMissEm]

I'm not sure if this would be suitable for our initial report nor for our first round of work. Unless there's some recommendation text we can come up with for this for the initial report that clarifies what AP already non-normatively specifies?

I'm going to label this as a future issue for now.