The annotation analysis simply annotates each SNP in a set for an overlap with the selected regulatory datasets. It also calculates the total number of overlaps for each SNP, which may serve as a crude estimate of SNP's functional impact. There are countless such databases and tools that do it better, such as FunciSNP, GWAVA, Annovar, RegulomeDB, just to name a few, and it is advised to use them for annotation of SNPs.
The output of the annotation analysis are available on a corresponding tab on the results page. For each SNP set, a subtab is created.