Skip to main content


 
- #unitoodailynews, #freesw, #softwarelibero, #software, #mobile, #laptop, #os Curbing Harassment with User Empowerment – Purism https://t.co/SmLChWYapo -<br>
Image/Photo<br>
- #Unitoo - Where #human #creativity plots the #possibilities of #technology -<br>
- #UnitooInc #News #Worldnews #tech #bot

Twitter: Unitoo on Twitter (Unitoo)

Harassment can be somewhat automated by writing a good algorithm which is self-learning. I.e. the message from a supposedly "harassing" user can be given a confidence score from 0-100. If a score is say, below 90, then the comment is sent to a human moderator. The suspect user can be added dynamically to a list (of users) that changes as the user's behavior changes. Thus as time goes on the suspect user will drop off the list if he behaves. If he starts up questionable activity again he gets on the list again. Very simple.

As time goes on this algorithm will learn better how to ID potentially harassing messages.

For example, before heuristic spam filters arrived or were very good, my spam email filter, around the time of 2001, was set to filter out any messages with 3 or more exclamation points. It worked REALLY well and had a very low false positive rate and a very low false negative rate. False positives were about 1 in 200 messages, until I told my step mom to stop using 3 !!! in a row.