In general, a digital twin consists of data from the physical model, typically measurement data, and data from the digital model, typically simulation data.[1] Both categories of data need to be stored in a memory efficient way that ensures the integrity of the dataset and allows the extraction of subsets of data with high performance and minimal memory requirements. In complex domains, such as Electromagnetic Capability (EMC), the data collected is likely to be diverse, variable and of large volume. These characteristics are properties of big data, which is the reason why a NoSQL database is a suitable technology for handling this data.[2]. In this work we have developed a generic data model for a document-based database, which performs particularly well with large scale object structures. So far, a comparable model has only been developed for graph-based databases.[4] In addition to managing the potentially large dataset of a digital twin, it is essential to ensure its validity. One way to automate the validity check is to add semantic information to the dataset and create a knowledge graph from it. We propose an approach to extract the semantic information from an existing dataset using our data model and thus creating the knowledge graph without human intervention. In addition to performing a validity check, the knowledge graph also has the potential to act as an expert system.[3]