Recognize and Prevent the Cyberbullying Conversation on Social Networks Using Machine Learning Technique

Abstract

In current days there was a lot of abused communication found in social media. A recent survey report confirmed that more than 80 percent of online social networks are having abused or vulgar communication on their user accounts. The process of threats or harassment of any user with the help of posting wrong/abused or vulgar messages using social media on the internet is known as Cyberbullying. These types of messages are mainly posted on user walls in order to harass teens, preteens other children by posting these types of offensive messages. Till now no application is providing a solution for this cyber content not to spread on social media, so this motivated me to design this current application for stopping vulgar communication in online social networks. In this proposed application, we mainly try to propose a new representation learning method to tackle this problem for identifying and stopping the abused messages not to communicate in online chat. Here we try to use well-known machine learning algorithms such as Support Vector Machine (SVM) for classifying the abused messages and normal messages and also we use Porter Stemming Algorithm to pre-process the text messages. This Porter Stemming is a well-known NLT Package(Natural Language Toolkit), which will divide the whole message into parts and then assign tokens for each and every individual word. Here we classify the cyber bulled conversation into five categories that are available in the literature like “HATE, VULGAR, OFFENSIVE, SEX, and VIOLENCE”.