The rise of social media has led to various issues related to online bullying. A serious drawback to social media is the limited ways that we can deal with cyberbullying as laws haven’t caught up with the new technology. Recently, researchers at King Saud University in Saudi Arabia have developed an innovative method to detect cyberbullying on Twitter with the use of deep learning known as OCDD.
Instead of feeding tweets to a classifier, as some other deep learning methods have tried, this system represents a tweet as a set of word vectors. Researchers across the world have been testing new ideas to reduce and manage cyberbullying, but many of those techniques come with limitations.
Various techniques currently in use, fail to consider user data including age and date of birth, which can be easily forged. The more features put into place to detect cyberbullying, the more muddled the feature extraction and selection phases become. In an effort to combat the current limitations of detecting cyberbullying, two researchers at King Saud University, Monirah A. Al-Ajlan and Mourad Ykhlef, developed the optimized Twitter cyberbullying detection (OCDD).
“Unlike prior work in this field, OCDD does not extract features from tweets and feed them to a classifier: Rather, it represents a tweet as a set of word vectors,” says the researchers in their published paper for IEEE Explore. “In this way, the semantics of words are preserved, and the feature extraction and selection phases can be eliminated.”
The students facilitated their approach based on training data and generated word embedding for individual words utilizing GloVe, described as, “an unsupervised learning algorithm that can obtain vector representations for words.”
In their published paper the researchers say, “OCDD advances the current state of cyberbullying detection by eliminating the hard task of feature extraction/selection and replacing it with word vectors, which capture the semantics of words, and CNN, which classifies tweets in a more intelligent way than traditional classification algorithms.”
In the testing phase, OCDD was rather successful. The method developed by the students has yet to be implemented or measured within cyberbullying detection contexts. Looking toward the future, researchers hope to adapt the technology to analyze text in Arabic.