You can certainly add inclusive filters. 
 Like: +<web root>/ViewPdf?artid=*

Project size makes incomplete mirror possible  
 Are there Article Link/Index pages? 
 Are article URLs predictible?  
It may be better to break this into many projects. 
Is there a simple way to verify completeness of mirror? (Do you know the file
size of the total articles?)

HTTrack follows links & downloads based on:
 domain/URL/file-name/file-type/file-size
 Are articles identifiable by these aspects? 
 What do the sites/articles looks like?