Introduction
Most applications handle implicit information, relying on knowledge about the context by the user and the application.
But that is not good enough for global semantic interoperability, as is the target of ISO 15926.
Discussion
Applications must have lightning-fast response times. For that reason shortcuts have to be made. As such that is not a problem when you know to use the application. But when information from that application is to be integrated with any other information about the process plant, the software can't possibly know the ins and outs of all the contributing applications.
Relational data base tables carry data that have 'something to do' with the subject of the table. The semantics are defined in the data dictionary (if at all) in a natural language. Writing crystal clear and complete definitions is difficult, so the mapping software has to understand the natural language used and the semantics of the definition.
A commonly made modeling mistake is to attribute stream data to the equipment containing that stream, for example "Pump B14-P-101 has a max. pressure of 12.4 barg", rather than representing the relationships in this information more explicitly as "Pump B14-P-101 contains Stream B14-S-P-101" and "Stream B14-S-P-101 has a max. pressure of 12.4 barg". Granted, that is more verbose, but it expresses the semantics in terms of ontologies, here two templates:
Any mapping from data in a proprietary format to ISO 15926-7/8 format hence does require proper data modeling, hopefully by good data modelers. They can be assisted by AI, see here .