Model-Driven Data Contracts for Digital Twin Services

2026-07-30

AI Summary

This study addresses the lack of systematic mechanisms in existing digital twin services to express and enforce data quality requirements—such as accuracy, completeness, and timeliness—at the model level during runtime, which undermines service reliability. To bridge this gap, the authors propose a contract-based approach to data quality management that formalizes a theory of data contracts, integrates it into the digital twin architecture, and introduces a domain-specific language (DSL) to support declarative specification and automated monitoring of these contracts. This work achieves, for the first time, a closed-loop integration of model-driven data contracts within digital twins, ensuring end-to-end data quality from the modeling phase through runtime execution. The proposed method significantly enhances the trustworthiness of downstream services, including simulation, what-if analysis, and machine learning–based prediction.

Abstract

Digital Twins (DT) integrate data from multiple sources. Models consume data and enable DT services such as simulations, what-if analyses, and ML-driven predictions. To ensure proper DT operation, data-driven services require data to exhibit traits such as reliability and high quality (including, e.g., accuracy, completeness, and timeliness). Yet, there is no systematic way to specify data requirements at the model level, and subsequently enact those specifications at runtime. To address this shortcoming, we propose an approach to contract-based quality management in DTs. We formally define a theory of such contracts, situate them architectually within DTs, and propose a domain-specific language to specify contracts. Our approach enables continuous data quality monitoring, thereby improving the reliability and quality of DT services.

Problem

• Digital Twin
• Data Quality
• Model-Driven
• Data Contracts
• Service Reliability

Innovation

• Data Contracts
• Digital Twin
• Model-Driven
• Data Quality
• Domain-Specific Language