An algorithm that detects misogynistic tweets
By Tanisha Jindal
Various social media platforms nowadays are flooded with harsh and irrelevant content against women. Excess of such content is uploaded on different social media platforms each coming day. But now it seems that there could be some relaxation as researchers have developed an algorithm that cuts through the rabble of millions of tweets and identifies misogynistic content over Twitter.
The researchers who developed this algorithm belong to the Queensland University of Technology (QUT) in Australia. The process to develop this sort of algorithm included mining of one million tweets and then the filtration of those tweets using three abusive keywords that are repeatedly used against women. Such practice made the process of acquiring an algorithm much quicker than the method of reading each tweet individually.
The major challenge was to classify if such tweets were whether a joke, said sarcastically or targeting anyone or a specific group. Regarding this matter, one of the co-authors Professor Richi Naya explained that any phrase such as "get back to the kitchen" will not be categorized as offensive or unjustified against women as it doesn't contain any abusive words. The machine learning classifier is developed in a way that it can certainly differentiate between such tweets.
Researchers found that the algorithm can identify misogynistic tweets with 75% accuracy in the testing executed by them. Thus, the researchers hope to see the tool getting used by various social media companies as it can support in recognizing racism, homophobia, and even abuse against disabled people.