Detection of Abusive Content Using Machine Learning and Deep Learning Methods

Abusive language refers to any expression that contains offensive terms in conversation, including social media platforms and networking websites such as Facebook or Twitter. As more users take to social media platforms that offer anonymity and easy access, the use of abusive content has increased rapidly. Therefore, there is a need to build systems that can identify offensive content efficiently to take appropriate actions quickly and swiftly. In the proposed work a model for the identification of abusive content from both text data as well as image data is presented. A web application for abusive content identification is developed using Django in Python environment. For analysis of abusive content in text data Natural language processing and different machine Learning algorithms are used while, for the detection of abusive data in images deep learning algorithm is implemented. The results attained illustrated that the proposed model detects abusive content with an accuracy of 82% using Logistic Regression classifier and convolution neural network.