Created a program to remove misogynistic messages

Researchers at Queensland University of Technology have created an algorithm to remove

misogynistic messages from Twitter. At the same time, the program can allegedly distinguish sarcasm from a real insult.

First, the researchers trained the algorithm ona million tweets. They then refined the dataset by searching for messages with three keywords: “whore”, “slut”, “rape”. The researchers also classified tweets based on whether they were misogynistic or not, based on the context of the messages. As a result, the algorithm learned to search for misogynistic publications.

The researchers noted that a kind of victorywas the moment when the program was able to label the phrase “get back to the kitchen” as misogynistic. This meant that the algorithm had learned to take context into account.

The authors of the development claim that the system is capable of detecting misogynistic tweets with 75 percent accuracy. The researchers hope their algorithm will help protect women online.