First, I am not sure of this answer; but I hope it will help you.
What you asked for, must be located in the following link : POS tag set does the parser use? (but unfortunately there are many broken links!!).
As they mentioned, you can find the tag set (you called labels) in the following file atb1-v4.1-taglist-conversion-to-PennPOS-forrelease.lisp. As I understand it, they map one or more sequence of English tags to one or more Arabic tag. for example:the mapping:
(DET+ADJ_COMP+NSUFF_MASC_DU_GEN DT+JJR)
means that we map the English sequence of tags (DET+ADJ_COMP+NSUFF_MASC_DU_GEN) to the Arabic tags (DT+JJR).
Regarding your question (المدرسة/DTNN), they mentioned that it consists of two tags (DT + NN) where DT= (الــ) (pronounced as 'Al') and NN = (Noun, singular or mass) see Penn Treebank .