We are participating in the challenge, and we would like to use the PDTB database (we have access to it). We were wondering if the scripts to prepare the data (as the one reported in the CNC paper) can be shared? Or at least the train/dev/test/ split?