ML-Based Translation Methods for Protocols and Data Formats

In order to exchange information between systems, the information must get encoded into a predefined data format, and it must be transferred in a protocol that the communicating parties have agreed upon. This works well if all parties follow the same protocol standard and use the same data description schemes. If systems use different data formats or protocols, then some sort of translation is required. Protocol and data format translation has been attempted previously through rule-based approaches, ontologies, and also by using machine learning (ML) techniques. Due to the current advances related to AI/ML methods, tools, and infrastructure, the accuracy and feasibility of 'translation' with ML-approaches improved significantly. This paper introduces a generic approach and methodology for translating data formats and protocols with ML-based methods and presents our initial results through JSON-XML and JSON-SenML translation. © 2023 IFIP.