Hello Kida,
Welcome to the Qiime2 forums!
This is a great question. First, you can choose a denoising/clustering pipeline like deblur or DADA2, then select a database later.
The database should match your amplicons, like Greengenes for 16S or Unite for ITS.
Some classifiers are customized for a region, like V4 or V1-V2 like you have.
If you want to use Greengenes2 for V1-V2, use the qiime greengenes2 non-v4-16s
action.
Or go for something simple and classic like classify-consensus-vsearch
Human stool is well-studied, so any database should work well!