System Evolution Analytics on Interconnected Entities in a System State Series

ABSTRACT This paper presents a System Evolution Analytics (SysEvolytics) model to capture system evolution information. We express evolving systems as a State Series , such that each state is preprocessed to generate temporal networks as an Evolution Representor (ER), which can be used for the analysis of evolving interconnected entities (or features) in the state series. For the System Evolution Mining , we used two types of pattern mining as follows: evolution rule mining and evolution subgraph mining. For the System Evolution Learning , we used interconnected entities in the ER to learn and recommend using System Neural Network . To generate knowledge and recommendation reports about system evolution, we applied SysEvolytics model on software, natural language, retail market, and movie genre. We discussed three SysEvolytics applications on Cloud services, Bitcoin, and Social media. The System “ilities” (changeability, stability, complexity, interpretability, and explainability) are defined to help stakeholders as follows: analyst, developer, tester, and maintainer.