ReviGO looks like it will cluster your GO terms by similar process, which I think is what you want to do. As a follow-up, if you have some basic programming knowledge, you could then write some scripts to parse your Blast2GO output file and get the list of genes associated with a specific GO term cluster you are interested in. I am assuming this is ultimately what you want to do.
Hope that helps!
Do you want to cluster GO terms themselves or actually cluster your sequences using functional similarity indicated by associated GO terms ?