you need to look at the site definition SPSPORTAL &
BLANKINTERNETCONTAINER in the WEBTEMPSPS.XML file (specifically the
configs SPSPORTAL#0 & BLANKINTERNETCONTAINER#0. Both of the
configuration notes point to a provisioning assembly. If you look at the
ProvisionData attribute (for example in BLANKINTERNETCONTAINER#0),
you'll see it points to xml\InternetBlank.xml.
Look at ..12\TEMPLATE\XML\InternetBlank.xml... the structure in that XML
file is what is generated when you create a new site collection using
the Publishing Portal template.
You can easily create your own site definition and specify your own
ProvisionData data. You'll find the schema for this file in the
..12\TEMPLATE\XML\PortalTemplate.xsd.