"Two questions from Adaboost"

Hi,

I'm new in rapidminer.

I want to use "adaboost"  to make my dataset more efficient  and then feed it to neural network for predicting the outliers. I have two questions:

1) (adaboost result contained 1 to 9 models with diffirent weights.) I need to know which of these models is more trusty and operative?

2) The output of adaboost is a model. How to select the best classified attributes from adaboost output model and feed it to neural network as a training set???

Thank you in advance