The different solutions that have been applied in the fight against cyberbullying have not worked. Or at least not yielded the desired fruits.
The good news is that inventors have not given up and the new cyberbullying detector is proof of this.
The cyberbullying detector program was developed by researchers at the University of Colorado Boulder (CU Boulder) to fight bullying on social media platforms.
The CU Boulders’ CyberSafety Research Center presented the computer program and app in recently published research.
If introduced into the market, this program could be a potential solution to the rising cases of cyberbullying in various parts of the world. Alternatively, it could pave way for the development of similar programs customized for different regions, cultures, and languages.
Cyberbullying cases have been increasing in Kenya. Women have especially been subjected to this form of online harassment and while the recently signed Computer and Cybercrimes Bill 2016 spells out stiff penalties for cyber crimes, the low reporting rate may spare guilty parties.
To date, some damage has already been done.
In early 2017, a woman identified as Brenda Akinyi is said to have committed suicide after a cyberbullying incident.
The 29-year old was allegedly bullied online by some members of the “Buyer Beware – Kenya (ORIGINAL)” group after posting about the defilement of her three-year-old daughter. It is alleged that some members started talking ill while others bashed and ridiculing her.
While the said Facebook group was in the eye of a storm for a while, the noise soon stopped. This is the norm in Kenya. Social media users make noise and leave the follow up to others. Always onto the next.
Brenda is just one among many victims of cyberbullying. Most of them are silent victims.
This silence and half-arse follow-ups may explain why we are yet to get a workable solution to cyber harassment. Perhaps until now?
How the program works
The cyberbullying program scans and analyzes social media data for abusive language and incidents of bullying. This includes patterns and incidents of online bullying.
This program works on both desktop and mobile: It is both a desktop computer application and an Android mobile app called BullyAlert.
The program targets public accounts. Private Instagram accounts are spared.
If the program detects abusive comments or characteristics of cyberbullying, it sends an alert to administrators who take it from there.
But, BullyAlert app can do more than that, it allows parents to monitor their children’s Instagram accounts. It achieves this by following individual Instagram posts.
At the moment, the cyberbullying detector is a functional prototype and is only compatible with Instagram. However, expansion plans are underway and it is set to get capabilities to analyze other social media platforms such as Facebook and Snapchat.
Not so fast though
We may have a problem. The kind that cannot be overlooked.
The concept is great, but some skeptics have brought out the issue of limits of machine learning. While artificial intelligence has come along way, we still have a long way to go. That may explain why some concerns have been raised about the effectiveness of the new program.
One of the concerns is on the definition of cyberbullying. The definition is quite subjective and varies from person to person and culture to culture. This means that the program would not be very effective unless it was customized for different cultures.
There’s also the concern that artificial intelligence is not adequate to correctly understand and contextualize interpersonal communication. Skeptics have argued that a computer program may not always accurately differentiate between abusive language and sarcasm on social media.
According to CU Boulder Today, a pilot test conducted showed that the program has a 70 percent accuracy.
Despite the glaring challenges, any tool that promises to help is always worth investigating in. We look forward to a possible rollout in various parts of the world and the development of even more innovative solutions.
Leave a Reply