Regex-Based Structuring of Audit Logs for Machine Learning–Driven Web Intrusion Detection in Golang Applications Dataset

Indonesia's rapid digital expansion has created significant cybersecurity vulnerabilities, including Golang applications. This research addresses the challenge of analyzing and collecting attack datasets targeting Golang applications, which pose substantial obstacles for security researchers attempting to understand attack patterns and develop effective defense mechanisms. We propose a systematic approach for gathering and analyzing web application attack data specifically targeting Golang-based systems, focusing on three attack vectors: Cross-Site Scripting (XSS), SQL injection (SQLI), and Server-Side Request Forgery (SSRF). Our methodology employs controlled testing environments to simulate realistic attack scenarios against purpose-built Golang web applications. The data collection process involves deploying vulnerable Golang applications with logging mechanisms to record attack attempts, successful exploitations, and system responses. We generated a dataset of 411 attack samples with 64 candidate features, covering SQLI, XSS, and SSRF attacks targeting Golang applications. Each attack instance includes request parameters, HTTP headers, response codes, payload characteristics, and exploitation outcomes. Evaluation using 12 baseline classifiers demonstrates that tree ensemble methods achieve 99% accuracy when utilizing the full feature set. Removal of categorical headers reduces balanced accuracy by up to 10%. This dataset contributes to the cybersecurity research community by offering real-world attack samples that can be utilized for developing and evaluating intrusion detection systems, web application firewalls, and automated security testing tools specifically designed for Golang environments