User behavior analytics from log data in cloud-native applications

As the use of cloud-native programs increases, it is important to be able to analyze user behavior using log data to address security, operational effectiveness, and user-driven customization. This review intends to accurately and thoroughly evaluate AI approaches that have been developed in the last 10 years for User Behavior Analytics (UBA) from log data. To do this, we review developments in machine learning, deep learning and hybrid approaches, and we present a systematic categorization of the approaches, summarize their experiments and findings, discuss challenges, and outline future work that could include privacy preserving UBA, cross-platform generalization, and real-time analytics. Through mapping the experience and forecasting the future, we hope to provide a historiographical reference for researchers and practioners aiming to deliver effective, responsible, and scalable UBA approaches in cloud environments.