Semantic transformation within message exchange between heterogenous systems provides a challenge for any enterprise. The primary reason for this challenge is that the required transformations are based on arbitrary codes that do not adhere to any grammatical or linguistic structure or rules. These codes are assigned to describe an entity or attribute such as order type or equipment ID. In any message exchange, the meaning of the data must be preserved. This is done today by applying translation lookups to simple cross-reference tables or files scattered across the computing landscape for each application use case. Such tables are maintained manually and can cause processing delays due to failed transactions that rely on semantic transformation values. The research presented in this dissertation introduces generic transformation patterns and applies these patterns to an underlying graph structure. By generalizing semantic transformations through patterns and graphs (Chapter 5), it is possible not only to determine a traversal path to find a corresponding code value, but to predict missing nodes through the use of graph embeddings (Chapter 7). The presented architecture is based on microservices. This allows for horizontal and vertical scalability and provides an abstraction to the transformation query such that various forms of queries can be submitted and semantically transformed without changes to the underlying traversal algorithms. The thesis demonstrates two scenarios. The first is a natural language scenario (Chapter 6), and the second is an event stream scenario. In the natural language scenario, a natural language query is provided, and the system returns the expected code-based results. This use case demonstrates the benefits of integrating NLP capabilities for users to search and find transformation codes via a web front-end. The second scenario demonstrates an IoT case where components of sensor data need to be enriched so that downstream data analytics applications can deliver the correct KPIs and dashboards. The value of this research is demonstrated by a cost analysis for failed transactions (Chapter 8) and by the benefits the presented semantic transformation architecture would provide to corporations. This research therefore underlines the importance of incorporating semantic transformations into organizations’ overall data architecture and strategy.