<![CDATA[This research is motivated by the increasing threat of cybercrime, particularly phishing attacks that exploit the rapid development of information technology to obtain sensitive user information, such as login credentials and personal data, through fraudulent websites or malicious links. One of the approaches that can be used to detect phishing attacks at an early stage is the analysis of suspicious URL patterns and structures [1]. Therefore, this study aims to design a web-based system capable of identifying phishing site patterns using a hybrid approach that combines rule-based methods and Large Language Models (LLMs). The rule-based method is utilized to extract indicators or digital traces from URL structures, including the use of IP addresses instead of domain names, the presence of the "@" symbol, abnormal URL lengths, sensitive keywords, URL shorteners, excessive numbers of subdomains, and hyphens in domain names [1], [2]. The results of this initial analysis are used as indicators of predictable phishing patterns. However, when the rule-based method cannot adequately identify suspicious indicators, the LLM performs advanced analysis to evaluate unpredictable phishing patterns and generate a more comprehensive final decision. The system is designed to produce two types of analysis results, namely rule-based detection results and LLM-based analysis results. In addition, the system provides data storage functionality and PDF report generation as a form of digital documentation. The results indicate that rule-based methods alone have limitations because they can only detect predefined indicators. Therefore, integrating LLMs improves the system’s ability to identify more complex phishing patterns and previously unseen threats that are not explicitly defined within the rule set.]]>
