Those IBM EBCDIC conversion tables are maintained by IBM. As Markus mentioned, they came from official sources at IBM. They weren't derived nor guessed. If they are out of date, someone from IBM would have to convert and contribute that data from the official source again.
In the meantime, you should consider transitioning to Unicode completely. It's unlikely to get a fast response in the meantime. Charset conversion is full of issues like these, and it's best to avoid the problem completely.
George