Core / Extensions

[1313ou]

Strictly speaking, this is not an issue but tentative proposals, open for discussion.
Here they go:

1 - It is appropriate to distinguish between core levels and extensions

2 - The core level should define
- the core contents of a word net
- possibly an extension mechanism, but not the extensions themselves

3 - The core level should deal only with stand-alone internal coherence and well-formedness. It excludes external references to external databases. Internal references should be checked.

4 - Extensions are permissible and should have their own namespace. They can deal with external reference. They are responsible for their own validation.

5 - Each extension should provide a strip-down mechanism that produces core-conformant data, in effect stripping down non-core data.
This can be easily done with each extension providing an XSLT transform script "to_core.xsl" that, when invoked, will filter non-core data away.
Invocation can then be triggered by simply including this in the XML header:
<?xml-stylesheet type="text/xsl" href="to_core.xsl"?> or passing the XSLT to DOM builders.

[jmccrae]

To clarify this, are you talking about extensions to the schema (e.g., for the modelling of pronunciation, etymology, morphology) that some wordnets have or is this related to "Core WordNet" (the 5000 most frequent senses of the English WordNet)?

[1313ou]

The first.