Very difficult to put simple rules like that in the world of cells and bio-molecules. Some transcription factors (TFs) have both activation and suppression activities. But the dual role could be brought by means of co-factors (e.g., ER alpha). Some TFs may have both domains (e.g., NF1/X - pubmed ID 9822643). Most activating transcription factors have a single activating domain. But the possibility of multiple activating domains in a TF cannot be ruled out (e.g., see pubmed ID: 9545321& 8525377).

Is there a website for prediction of activation domain in a transcription factor?

I agree with Kshitish Acharya, Its hard to put a rule of 1:1 for domain vs factor! In fact repressive factors are known to harbor multiple domains - each acting in different for repression - some enzymatically versus some via DNA binding and others via scaffolding action of other accessory factors (eg. DNMT1). But if you want to model this, one can start with a safe assumption that for DIRECT effects, if a factor has ONE DNA binding domain (about 1 turn of a helix, refer RecA/Rad51 studies) - it can correspond to one activation domain. The larger the protein becomes the more complicated it can get due to message cross talk!