March 20, 2018 | Written by: Grant Gelven
Categorized: AI/Watson | Developer | Social Service
Share this post:
Cyberbullying is a critical issue to the health and mental health of our children and teens. Eighty-seven percent of youth have witnessed cyberbullying. Nearly two-thirds of students who have experienced cyberbullying stated that it affected their ability to learn and feel safe at school.
At Identity Guard, we began looking into how our monitoring services, empowered by IBM Watson technologies, could be applied to this issue in 2016. We reached out to academics and experts to learn more about cyberbullying and its consequences, and to ascertain how we might design an effective solution to combat it.
In this process, we worked with the Megan Meier Foundation, and in hearing Tina Meier’s story about how her daughter took her own life as a result of cyberbullying, we knew we had to get an effective solution into the market as quickly as possible.
Developing a solution to address cyberbullying
Through our consultations with the Megan Meier Foundation, we developed a good understanding of what kids and parents go through in cases of cyberbullying. We knew that whatever we designed had to protect teens’ privacy, educate parents and equip both with tools on how to best respond.
We developed an additional feature that can be added to an existing suite of our services, where parents can add their kids to watch lists for social media monitoring. Based on advice from experts, we identified the networks with the highest risk of cyberbullying and have focused on these—Facebook, Twitter, Instagram, Tumblr and Google+.
Once parents identify which networks to follow, and their kids provide the necessary login credentials, the solution then monitors the feeds every few hours. If potential threatening material is identified, the parents are sent an alert, along with resources that help guide them on how to respond.
It’s a white glove, end-to-end service, so that we can help ensure kids are able to engage in social media safely, and if problems do occur, parents can respond constructively, versus out of panic.
Fighting cyberbullying with IBM Watson Technologies
The ingenuity, and differentiation, of our solution is that while kids’ social media networks are monitored, their privacy is protected—which is made possible by the power of artificial intelligence (AI).
With IBM Watson technologies that enable natural language processing (NLP) and natural language classification (NLC) we have developed an intelligent system that can extract, filter, understand and categorize the content of our user’s social media feeds. This provides a rich set of meta-data that can be used to determine if a post, comment, or tweet may be an instance of cyberbullying.
Throughout our development process, we used real-word examples of toxic, threatening, and aggressive language mined from a myriad of data sources to train individual Watson APIs to understand what makes these types of posts unique. Once we had individual Watson models that could detect these specific types of language, we stacked these models in an ensemble to produce a meta-algorithm capable of detecting many different types of harassing posts with a high degree of recall.
High recall is key in this instance because we wanted to provide a service that parents can trust, ensuring that when an instance of cyberbullying occurs, we will find it. This has the added benefit that even minor occurrences will be detected, thus allowing the parents to address the problem before it becomes a bigger issue.
If a problem of potential cyberbullying is identified, and only then, parents are sent an alert along with a snippet of the conversation, containing pertinent dates and times. The parent is asked if this is a potential threat, and if they confirm, they are provided with resources that include being connected to the Megan Meier Foundation, which offers expertise on relevant laws, school guidelines, a checklist for response, and an ear to listen to the parents’ concerns.
We are proud of what we have built because parents now have a tool that they can rely on to monitor for threats to their children while simultaneously maintaining a degree of privacy for them.
Helping stop cyberbullying from falling through the cracks
Due to the nature of cyberbullying, it can often fall through the cracks. Police don’t want to engage unless it’s a crime; teachers hesitate because it takes place outside of the school; and parents have a hard time keeping up with technology, as apps rapidly come and go.
With our cyberbullying service, Identity Guard, along with the Megan Meier Foundation, can provide detection, education, support and resolution. While it will initially be offered as an additional feature to an existing suite of services—such as monitoring of credit, dark web and social media for adults—we envision that one day it will be a stand-alone solution. Our goal throughout this work is to help stop cyberbullying, and in the interim, to equip parents with the necessary tools to appropriately respond when it does occur.