Abstract

Recent advances in data science have made available many URL-analysis–based detection and machine learning algorithms, which are being leveraged for phishing detection. It is necessary to measure the efficacy of each technique so as to select an efficient phishing detection technique for integrating into a system. There have been a number of studies stated so far in this related field to assist with this task. While most of the studies have covered all the approaches, only a few have concentrated their study on machine learning–based techniques. Additionally, the parameters such as true-positive, false-positive, true-negative, and false-negative rates have been analyzed in most of the studies in these related articles. In order to achieve new objectives in the field of security, the study on analysis of phishing detection techniques needs to be expanded. In contrast to state-of-the-art methods, our research scrutinizes the different mechanisms and taxonomy used in each machine learning–based detection technique succeeded by an upper bound computing time.

摘要

数据科学的最新进展提供了许多基于 URL 分析的检测和机器学习算法，这些算法正被用于网络钓鱼检测。有必要衡量每种技术的功效，以便选择一种有效的网络钓鱼检测技术以集成到系统中。迄今为止，在该相关领域已经有许多研究可以帮助完成这项任务。虽然大多数研究涵盖了所有方法，但只有少数研究集中在基于机器学习的技术上。此外，这些相关文章中的大多数研究都分析了真阳性率、假阳性率、真阴性率和假阴性率等参数。为了实现安全领域的新目标，需要扩大对网络钓鱼检测技术分析的研究。与最先进的方法相比，我们的研究仔细审查了在每个基于机器学习的检测技术中使用的不同机制和分类法，这些技术在上限计算时间上取得了成功。