New Yahoo AI Can Detect Online Abuse?

Online harassment is a serious, but tough problem to try and solve. 

A team at Yahoo recently developed an algorithm that claims to be able to automatically identify hateful speech. The tool uses deep learning to detect abusive keywords, punctuation that were typically found in hateful comments, and syntactic clues found in several thousand comments on Yahoo’s websites.

Researchers also using something called “word embedding,” which is the process of utilizing vectors to map out words and phrases often used in natural language processing for machines. 


According to Wired, the database will soon be released online on Yahoo Webscope, which would open it up to be used by other experts.

This is really an awesome step forward as the number of users online at any given time is huge and it is still growing.
Almost all of them have some or the other kind of interaction which utilizes chat and other forms of written (typed) comminucation
This is where the Artificial intelligence will come into play and hopefully be useful for the coming generations.

The study does take into account some of these issues. Spotting certain keywords is only part of it. An algorithm can’t detect sarcasm, nor can it keep up with constantly changing internet language and slang. More technically, a machine can have issues detecting hate across multiple sentences. Researchers wrote,


Social networks nowadays host millions of chats/comments online and is one area where this can be implemented and hence ease out the task of moderators who now have to kind of achieve this manually till a certain extent.

This is just the beginning, be ready for a lot more surprises and revelations in technology!