Handling "Unknown" variables in Enterprise Miner

[kewldude]

I'm not sure if this is the right forum, but I have just started using Enterprise Miner in my Predictive Analytics subject, so I thought of posting my problem here. 

You see I have a dataset that has some categorical cariables which has a value "unknown". This is not blank or missing per se, but an actual category level of a variable (i.e. marital status variable with single, married, divorced, unknown as the possible levels of the variable).

Going through the basic Enteprise Miner tutorial, I had a look at the Replacement and Impute node, but I'm not sure how to go about it.

What I wanted to do is to replace all these unknown values with something like the most frequent marital status value that was used in the dataset. 

Care to nudge me in the right direction?

[Reeza]

Use a decision tree to first predict this value as part of your model. Then use those predicted values in your model further on. 

[kewldude]

Just wondering if anyone can provide a quick example that I can use as a basis?