Detection Framework for Content Based Cybercrime in Online Social Networks

Abstract

The recent development of social media poses new challenges to the research community in analyzing online interactions among people. Social networking sites offer great opportunities for connecting people with each other, but also increase the vulnerability of young people to undesirable phenomena, such as content-based cybercrime. This may cause many serious and negative impacts on a person’s life and even lead to committing suicide. Cybercrime has emerged as a money-driven industry with malicious intent towards online social networks. Cyber-criminals aim to manipulate vulnerable areas in cyber-space by playing on human understanding and making a profit. They threaten minors, especially adolescents, who are not adequately overseen whilst online. In the recent past, the issues of Content-based Cybercrime have gained considerable attention. Social media providers seek for accurate and efficient way of recognizing offensive content for shielding their users. Content-based Cybercrime detection is one of the conspicuous area of data mining that deals with the recognition and examination of bully contents usually presented in social media. The current work emphasizes on cyberbullying, one of the prominent problems that arose due to the increasing fame of social network and its fast acceptance in our day-to-day survives. The social network provides a convenient platform for the cyber predators to bully their preys especially targeting young youth. In severe cases, the victims have attempted suicide due to humiliation, insult, and hostile messages left by the predators. xv To address this issue, there is an urgent need for a robust content based cybercrime detection framework. This thesis proposes three techniques for efficient detection of content-based cybercrime in online social networks. First one, cuckoo inspired SVM approach, aims to concurrently optimize the parameters and feature selection with a target to build the quality of SVM. This chapter proposes a novel hybrid model that is the integration of Cuckoo Search and SVM, for feature selection and parameter optimization for efficiently solving the problem of content-based cybercrime detection. In second approach, multiconfiguration detection technique, has been proposed to explore possible combinations of various preprocessing, feature selection and classification methodologies using the cuckoo search metaheuristic approach. This approach seeks to improve the performance of content based cybercrime detection system. In third approach, a novel cuckoo inspired stacking ensemble framework has been proposed that is the integration of cuckoo search and several machine learning models. The proposed framework automatically seeks for near-optimal combinations of classification techniques along with their tuning parameters for efficiently solving the detection problem of content-based cybercrime in multimedia platforms. The performance of the proposed approaches has been evaluated by testing on four different datasets obtained from Twitter, ASKfm and FormSpring to identify bully terms. The results of the proposed approaches demonstrate significant improvement in the performance of classification on all the datasets in comparison to recent existing models. The experimental results demonstrate the high efficiency and effectiveness of the proposed approaches. These approaches outperformed other recent techniques on all the datasets, giving high predictive recall value via 10-fold crossvalidation