You can find out about the corpus design on http://corpus.amiproject.org/documentations/ami-corpus-scenario-explained - the same four people participate in all four scenario meetings for the same series (e.g., ES2002a-d), but it's then different subjects for e.g. ES2009a-d.

To map from A,B,C,D (NXT agent designators) to actual people, you have to use the meetings corpus resource (meetings.xml), and the participants corpus resource if you want to know about the people.  You can do this using NXT if you want - here's an example.

java FunctionQuery -o ES2002a -c AMI-metadata.xml -q '($s speaker)($m meeting)($p participant):($m^$s) & (ID($p)=$s@global_name)' -atts '$m@observation'  '$s@nxt_agent'  '$s@global_name'  '$p@sex'

Corpus resources are packaged with the annotations.