Do you have an actual use case for this?
Seems to me that doing this sort of mapping you are changing the original context of the captured data.
Consider this. When you migrate/transform an XML document (schema constrained) into another XML document (different schema constraints), then you are potentially (very likely) changing the original intent of the information. If the two schemas have different constraints, whether those are ontological, temporal or syntactic, then the information no longer conveys the original intent.
For meaningful analysis, this is a dangerous step. Now if you do this multiple times you are headed down a road of confusion and misleading data.
Consider the case of where data is recorded for blood pressure and used to help trigger a warning for hypertension.
A schema is designed to establish constraints based on a clinical guideline. When a new clinical guideline is published, do you migrate the existing instances to match the new schema? If you do, what happens when you end up in court over a misdiagnosis? You will have likely lost the original context that applied to those measurements when the data was initially recorded. [1]
In a mostly free text scenario where you are using a schema for formatting purposes, this may well be okay.
The primary reasons for using a schema is to establish a coherent structure and setting constraints. "Validating against a schema is an important component of quality assurance." [2]
To answer your question, no I haven't. Nor can I imagine a situation where I would. I look forward to hearing use cases where this is being done.
