> Thank you. I understand that the operation mode is that the filter sends to>
> the monitor a score answer with the refer atribute and the unknown score,
> then the monitor invokes the heavy filter, isn't it? Is a CONT from the
> filter to the monitor needed then? I assume not.
I suppose that the CONT message is still necessary because the SCORE message is processed only from the DM; the monitor recieves it and redirect it to the DM without exmining its content. So if you want to activate your heavy filter you should tell to the monitor that you want it, using the CONT message.
This is my opinion, but other NLP filter developers (ILC, USF) can explain better the way they work with the monitor.