Machine Learning Approach Based on Hybrid Features for Detection of Phishing URLs

Phishing attacks are one of the most widespread problems over the internet. A lot of internet users fall into the hands of attackers every day which accounts into millions of dollars of fraud around the globe every day. The availability of the internet among people who don't have the knowledge of cyber-attacks adds more to this problem. Thus, there is a need to employ intelligent algorithms to solve these serious problems. In this paper, we present different ways in which phishing URLs can be detected using machine learning algorithms. The URL based features as well as network-based features were used to feed to the machine learning classifiers. Similarly, other features that might add relevance to our problem are also discussed. The unbalanced dataset is made balanced using various oversampling and undersampling techniques and the performance for the various machine learning algorithms is evaluated for the dataset. The evaluation shows that the machine learning algorithms can show promising results in terms of precision, recall, f- score, and ROC AUC.