Why use inverse of AsShotNeutral to white balance?

My question is the same as the subject. I want to know why should inverse of AsShotNeutral tag be used to get the white balance gains to white balance the RAW image? The DNG specification does not mention to take the inverse? Is it that I am missing anything or I am totally way off the truth? Please guide. 

For example my DNG file reads

AsShotNeutral = [0.4697, 1, 0.6465] ;

Only when I take an inverse and then apply the resultant white balance gains (by masking according to the RAW bayer pattern) I get the correct results. Why is that? 
I am following this document to achieve a minimal RAW processing.