A Machine Learning System to Classify Cybercrime

Abstract

The rapid enhancements in the field of computer systems and networks have provided a new way for unethical activities such as cybercrime public safety, security, and global economy are all under the threat of the cybercrime activities. Aim of the proposed work is to classify cybercrimes into cyberbullying and IP fraud analysis. Through cybercrime events providing a brief scrutiny about the cybercrime activities, their relative fundamentals and offering a pairing schema are the prime objectives of this work. The cybercrime incidents are classified based on their ideal characteristics using random forest machine learning algorithm. K-means clustering applied on the Twitter data is collected from Kaggle database. The optimal features like user ID, sign-up time, source, browser, sex, age, and IP address are extracted to train the machine. To analyze the performance of the proposed algorithm experimentation is carried out on total of 151,113 datasets. Compared to existing methods, the proposed method outperforms with an accuracy of 97%. Novelty Utilization of random forest algorithm for the classification of resulting cybercrimes helps to efficiently overcome the problem of regression. Different age groups form the basis for identifying the various offenses in the proposed work.