I was involved in some studies to capture an XML business language semantics using semantic Web technology such as OWL and I think things to consider when semantically analysing your airplane document would be the implicit relationships, their directions and the semantic labels applicable to those relationships. So

Duration and speed are properties of the flight, because by putting flight as a wrapper around both duration and speed you are making an implied child-parent between duration and flight and between speed and flight, and a sibling-sibling relationship between speed and duration. Likewise by making hours an attribute of duration you imply a relationship. The relationships are directed. Flight is not a property of duration in your implied semantics but it could be because a duration could possess a duration-type which could be flight. Another semantics would be to have duration as parent element and type=“flight” and speed= 500 kph as attributes or child elements. The two semantics would be equivalent but not identical. Then there is inheritance, aggregation, ownership - labels of the relationships to consider. That is just the beginning, though, because semantics is about meanings and you then might define what your document means by flight, duration, units, speed, kph. Here you would fit your names of elements and attributes into an ontology of the domain semantics. You might consider the codifications such as kph and hours. Plus the context such as commercial, supply-chain or military air travel.

So what you are describing as semantics is not the same. What you are calling semantics, I would call calculation model.