Dynamic Property Preservation in AIoT: A Machine Learning Approach for Data-Efficient Model Transformation

Abstract

Model-driven development (MDD) in the Artificial Intelligence of Things (AIoT) domain faces significant challenges in ensuring the consistency and preservation of model properties during transformations, often leading to system inconsistencies. This research introduces the Property Preservation Framework (PPF), a novel approach fortified with a Markov chain methodology, specifically designed to address these challenges. The PPF integrates formal procedures and constraint-checking methods to systematically validate and preserve model properties, thereby enhancing the reliability of AIoT systems. Through empirical evaluations, the framework has demonstrated its ability to efficiently determine and verify model characteristics at various transformation stages, significantly reducing the incidence of property violations. The results indicate that the PPF not only ensures overall consistency and reliability but also optimizes resource allocation, thereby enhancing data efficiency during the property validation and preservation processes. These advancements make a substantial contribution to the domain of MDD, providing developers with the methodology to execute model transformations that accurately reflect system requirements and behaviors in AIoT ecosystems. The findings underscore the potential of the PPF to revolutionize AIoT development by ensuring high-quality, dependable, and efficient modeling outcomes.

Keywords

• Model-driven development
• model transformation
• property preservation