Phishing is one of the most frequently occurring forms of cybercrime that Internet users face and represents a violation of cybersecurity principles. Phishing is a fraudulent attack that is performed over the Internet with the purpose of obtaining and using without authorization the sensitive information of Internet users, such as usernames, passwords, credit card details, and bank account information. Some widely used phishing attempts involve using email spoofing or instant messaging, aiming to convince a victim to visit the spoofed websites, which will result in obtaining the victim’s information. In this work, we identify and analyze the most important features needed to detect the spoofed websites in virtue of two new feature selection techniques. The first proposed feature selection technique uses underlying feature selection methods that vote on each feature, and if such methods agree on a specific feature, that feature is selected. The second feature selection technique also uses underlying feature selection methods that vote on each feature, and if the majority vote on a specific feature, the feature is selected. We also propose a phishing detection technique based on both AdaBoost and LightGBM ensemble methods to detect the spoofed websites. The proposed method achieves a very high accuracy compared to that of the existing methods.

网络钓鱼是互联网用户面临的最常见的网络犯罪形式之一，它违反了网络安全原则。网络钓鱼是一种欺诈性攻击，它是通过Internet执行的，目的是未经授权获取和使用Internet用户的敏感信息，例如用户名，密码，信用卡详细信息和银行帐户信息。一些广泛使用的网络钓鱼尝试包括使用电子邮件欺骗或即时消息传递，目的是诱使受害者访问受欺骗的网站，这将导致获得受害者的信息。在这项工作中，我们借助两种新的特征选择技术来识别和分析检测欺骗性网站所需的最重要特征。首先提出的特征选择技术使用对每个特征投票的基础特征选择方法，并且如果这样的方法在特定特征上达成共识，则选择该特征。第二种特征选择技术还使用对每个特征进行投票的基础特征选择方法，如果大多数人对特定特征进行投票，则选择该特征。我们还提出了一种基于AdaBoost和LightGBM集成方法的网络钓鱼检测技术，以检测被欺骗的网站。与现有方法相比，所提出的方法实现了非常高的精度。该功能已选中。我们还提出了一种基于AdaBoost和LightGBM集成方法的网络钓鱼检测技术，以检测被欺骗的网站。与现有方法相比，所提出的方法实现了非常高的精度。该功能已选中。我们还提出了一种基于AdaBoost和LightGBM集成方法的网络钓鱼检测技术，以检测被欺骗的网站。与现有方法相比，所提出的方法实现了非常高的精度。