Data is a globally recognized asset, with modern vehicles serving as significant sources of it. However, to effectively support decision-making, data must undergo preparation. Traditionally, automotive data and software architectures have been application-centric, leading to repetitive data engineering processes due to a lack of standards and semantic data models. This has resulted in isolated data models tailored to specific use cases, with no standard semantic model available to address the diverse dynamic properties of vehicles and the outcomes from applications that process them. These properties generate data streams that impose a one-pass constraint on architecture, necessitating immediate processing, storage, or forwarding of data to avoid permanent loss. Given the limited computational, storage, and bandwidth resources in vehicles, it is essential to manage data streams effectively for optimal utilization. This research explores designing an artifact that interprets vehicle data streams and conveys their meaning in a comprehensible manner, yielding instant, actionable, and reusable insights within vehicle architecture. Utilizing the Design Science Research methodology, it proposes an approach to work with vehicle data streams at a higher abstraction level than raw sensor data, featuring a semantic model, a continuous data aggregation process, a semantic annotation step for context, and a knowledge formalization step to
