Augmenting Digital Ecosystem Resilience Through Human-Centric Cybersecurity Solutions

As personal, social, and economic activities increasingly shift online, the underlying digital ecosystems are becoming more complex and susceptible to cyberattacks. While conventional cybersecurity solutions often target technical threats, they frequently overlook the human element. Human perceptions, behaviours, and decisions significantly affect system security, yet most current approaches fail to address these factors adequately. This paper fills that gap by introducing a system that interprets human-related cybersecurity data. It integrates multiple data types, including user activity logs and behavioural indicators, with a structured classification method to assess user-related risks more effectively. A reference model is also proposed in this work, combining zero-trust security (where no user or device is inherently trusted) with an adaptive trust framework that responds to individual user profiles. The system is evaluated against established machine learning methods, such as support vector machines and random forests, using a publicly available dataset that simulates both benign and malicious activity. We assess its performance in both risk classification and dimensionality reduction, reporting standard metrics such as precision, recall, and F1-score. Results indicate an 11% improvement over baseline models. Beyond technical validation, we conducted a user study to evaluate usability and user sentiment. Participants answered rating-scale and open-ended questions. Quantitative feedback highlighted improved satisfaction and usability, while qualitative responses were analysed for sentiment, positive, neutral, or negative. A 15% increase in agreement with user-friendly security practices was observed. These findings suggest that integrating human-context data with adaptive security frameworks enhances both technical efficacy and user alignment in cybersecurity systems.