Phishing Detection Dataset: Feature Engineering and Selection

The Internet has become an inevitable part of our lives. According the statistics two-thirds of the total world’s population is on Internet today. In recent decades, some unpredicted situations, such as COVID-19, have led to an increase usage of Internet services, not only for entertainment and work, but also for education and commerce. Many people have started blindly trusting and sharing personal information and credentials. This opens a bottleneck in the network security. Networks threats and attacks have been around for as long as networks have existed, and they evolve with the network progression. For this reason, the defense of network resources, critical data and personal information have to be addressed and constantly updated in order to face the modern attacks. In this paper authors address sophisticated network attacks, which are based on machine learning (ML) techniques and artificial intelligence. The fight against network offense based on ML and AI require an adequate response which utilize the same technologies. Developing defense mechanisms by using ML approach can provide more robust and effective protection against these complex and dynamically evolving threats. On this way attack can be predicted and thus prevented in real-time using the same techniques attackers employ. This paper focus on phishing detection and defense, based on machine learning. The success of any ML models is strongly dictated by the input data pass through the model during training. Therefore, an enhanced dataset for ML model construction is proposed.