Software Detects and Deters Cyberbullying
Kids who are bullied over the Internet often feel alone, which is a dangerous place to be. Isolation, fear and depression can drive kids to take extreme measures, even suicide. Karthik Dinakar from the Massachusetts Institute of Technology has been working on a project with her colleagues that could help reduce the feelings of loneliness experienced by victims of cyberbullying. The researchers are also hoping to developing software that can identify cyberbullying at its source.
Dinakar's team developed an algorithm that that analyzes words and phrases from the website, A Thin Line, which was developed by MTV as a way to empower young people to anonymously identify, respond to and stop the spread of digital abuse. The site gives teens the opportunity to vent their frustrations and give each other advice on how to deal with them.
The algorithm analyzed the words and phrases from 5,500 posts from the site and placed each one into one or more of 30 themes, such as relationships, sexting or sharing embarrassing photos. New Scientist gives this example: If a post contains the words or fragments “boyfr,” ”trust,” ”cheat,” ”break” or “upset,” then it could go into the relationship category. The software then matched up simliar messages, letting those who posted the problems know that they weren't alone.
Dinakar is also working on software that could identify cyberbullies at the source. Drawing from an open-sourced database called ConceptNet, which allows computers to identify words and their relationships with each other (i.e. dress/girl). For example, calling a girl a cow would pop up on the software’s radar as a negative comment, it knows that a cow is an animal, not a human. The software could be used in social networks to deter bullies by identifying phrases that could be offensive and then offer help to the recipient of the bullying.
Eventually, Dinakar wants to combine the algorithm analyzing words from A Thin Line with the ConceptNet database to create a cyberbully-deterring system. While it may not get to the root of the problem, it may be a step in the right direction.
via The New Scientist
Credit: Victor Habbick Visions/Science Photo Library/Corbis