Explainable machine learning for phishing feature detection

Abstract

Phishing is a very dangerous security threat that affects individuals as well as companies and organizations. To fight the risks associated with this threat, it is important to detect phishing websites in a timely manner. Machine learning models work well for this purpose as they can predict phishing cases, using information on the underlying websites. In this paper, we contribute to the research on the detection of phishing websites by proposing an explainable machine learning model that can provide not only accurate predictions of phishing, but also explanations of which features are most likely associated with phishing websites. To this aim, we propose a novel feature selection model based on Lorenz Zonoids, the multidimensional extension of Gini coefficient. We illustrate our proposal on a real dataset that contains features of both phishing and legitimate websites.

Developing a Biblical Solution Model for Mitigating Phishing Risks Among Internet Banking Users in Nigeria: The Initial Investigation.

“…Phishing Attacks (PAs) stand out among the many types of cyber threats because they use social engineering techniques to prey on users' psychological weaknesses and technological deficiencies Sameen et al, 2020). Not only do such fraudulent activities result in substantial financial losses, but they also erode the trust that underpins digital transactions (Calzarossa et al, 2023).…”

“…A compelling illustration of the severity of this issue can be found in the data provided by the Anti-Phishing Working Group (APWG, 2022), which highlights the financial sector, including banks, as the most heavily impacted by phishing attacks in both 2019 and 2022 (Calzarossa et al, 2023).…”

The ENBIS‐22 quality and reliability engineering international special issue

“…Misai et al 5 mix parametric, non-parametric, and machine learning inference methods to optimize a maintenance policy for a partially observed multi-component process. The following two articles by Calzarossa et al 6 and Rotari et al 7 share the same objective of variable selection when using random forest for classification or regression. The paper by Calzarossa et al 6 focuses on the application of detecting phishing sites based on URL characteristics; variables are selected in two stages: first by a univariate exploratory analysis, then by a Lorenz Zonoid-based selection procedure.…”

“…The following two articles by Calzarossa et al 6 and Rotari et al 7 share the same objective of variable selection when using random forest for classification or regression. The paper by Calzarossa et al 6 focuses on the application of detecting phishing sites based on URL characteristics; variables are selected in two stages: first by a univariate exploratory analysis, then by a Lorenz Zonoid-based selection procedure. Although, at the end, the method is applied to an additive manufacturing case, Rotari et al 7 is more methodological.…”