In cases of XML import where primary node type information is unavailable, either because it is skipped (see 7.3.3 Respecting Property Semantics, immediately above) or because it is not available in the first place (as is the case on document view import of arbitrary XML), the implementation must determine an appropriate node type to assign to each newly created node. This specification does not attempt to define or restrict how this is done. However, simply for the sake of illustration, some of the possibilities include:
Making all imported nodes nt:unstructured (obviously this will only work if the implementation in question supports that node type).
Dynamically creating node types appropriate to the incoming nodes. This approach might be suitable in cases where the incoming structures all fall into a few well defined and easily recognized patterns.
Use node types created according to structure information provided to the repository from an external source such as a schema.