SecureCorpus

This dataset abstract describes SecureCorpus, a specialized corpus developed for training and evaluating artificial intelligence models in forensic malware analysis. It is designed to interpret Sysmon event logs to identify and analyze malicious software behavior. The dataset was created using a hybrid methodology that combines data from controlled attack simulations with synthetically augmented samples. This curated approach ensures a diverse and comprehensive collection of event logs. The primary purpose of SecureCorpus is to enable the fine-tuning of Small Language Models or SLMs for on-device analysis on resource-constrained end devices. The dataset provides the necessary foundation for training models like SecureSLM to perform detailed threat detection and behavioral analysis efficiently. It facilitates the development of compact AI agents that can provide in-depth breakdowns of suspicious activities without relying on cloud resources. The use of this dataset helps to advance scalable security solutions that prioritize on-device interpretability and detection accuracy.