Development of new approaches in detecting social engineering attacks using deep learning methods and models

As a result of the developments in today's digital world, ensuring the security of the data shared by users on social media increases the importance of information security. In digital environments where the necessary security is not provided, all kinds of data of users become the open target of malicious people who are considered as attackers. All kinds of data of users, especially personal data and business-related data, can be seized by attackers. Social engineering attacks, which are one of the most common methods used to capture data, pose a great danger in terms of information security and data security. While keeping users at least as safe in digital environments as in physical environments is one of the most important problems of today, ensuring this security is primarily based on detecting and preventing these attacks. For this reason, the most common types of social engineering attacks must first be closely recognized. Then digital methods must be developed to detect vulnerabilities in these attacks. Within the scope of the thesis study, an innovative approach has been proposed to detect smishing attacks, the most common social engineering attack, to protect users' data security. It is aimed to protect users from these attacks by detecting these attacks with Natural Language Processing techniques. Smishing attacks were detected with the Hybrid CNN-BiLSTM method using Natural Language Processing techniques. Model performance was achieved as 0.998. Then, SMS phishing detection was performed using the Hybrid CNN – GRU method. The success rate of the detection performed with this method was obtained as 0.996. Finally, hate speech was detected with GRU on short texts using Natural Language Processing techniques and Twitter data. A performance of 0.998 was achieved with the GRU model. The aim of this thesis is to detect social engineering attacks using Natural Language Processing techniques and to contribute to the literature in this field.