MODELS FOR AUTOMATIC DETECTION OF CYBERBULLYING ON ONLINE SOCIAL MEDIA

Abstract

With time, use of the internet has become very common among people, and thus the rise of internet usage has given birth to a problem of cyberbullying. Cyberbullying can have a serious impact on the psychological health of the person who is the victim of it. Hence, detection of cyberbullying is required on the internet or social media. Much research has been done in the field of detection of cyberbullying. Machine learning is one of the approaches that can be used for automatic cyberbullying detection on online social media. Study on some of the papers related to cyberbullying along with some of the NLP techniques and different models used for cyberbullying detection tasks has been done . The graph, which is based on the papers reviewed, shows that the tf-idf is mostly used either directly or with a combination of other techniques for feature extraction in cyberbullying detection using machine learning. Five different ML models are used for classification of tweet data as bully or non bully text. Performance calculations of the model are done using accuracy and confusion matrix. Tf-IDF feature extraction technique is used to convert text into vector form. Random Forest model performs best followed by LR model. All the accuracy of the models are shown graphically also.