That sounds like a challenge Maruis. ^^ I'm sure it's technically possible to parse an PMML model in and convert it into the RapidMiner XML model format, save this and then read it in using the ReadModel operator. The two formats do have some similarities (at least for decision tree anyway) Although an 80 line PMML becomes nearly 400 lines of RapidMiner model XML. I imagine it could be done via various XML templates for a few select models which then are populated once the PMML has been read into RM. Not exactly sure the benefit of doing this except in that RapidMiner (even the server version) doesn't support PMML imports as a scoring engine, yet.
Contributor I
:-[ 
Answers
In version 5 there is the Read XML operator (it is available in version 6 but requires an additional licence).
You can then use the Generate Extract operator and some XPath to extract the things you want.
In fact there is an example here http://rapid-i.com/rapidforum/index.php/topic,3599.0.html.
regards
Andrew
Best regards,
Marius
I'm sure it's technically possible to parse an PMML model in and convert it into the RapidMiner XML model format, save this and then read it in using the ReadModel operator. The two formats do have some similarities (at least for decision tree anyway) Although an 80 line PMML becomes nearly 400 lines of RapidMiner model XML. I imagine it could be done via various XML templates for a few select models which then are populated once the PMML has been read into RM. Not exactly sure the benefit of doing this except in that RapidMiner (even the server version) doesn't support PMML imports as a scoring engine, yet.
Back to the original poster though, Andrew Chisholm put together a blog post advising on how to read PMML models with RapidMiner, perhaps this could be a starting point? http://rapidminernotes.blogspot.com/2011/05/how-to-read-pmml-file-to-determine.html
link broken