I understood that the difference in probability distribution was examined according to the presence or absence of context in the CAD method.
However, for the summary task, the context must always be provided.
It's unclear from your paper how this was contrasted.
Can you tell me how you did it if you didn't give context when evaluating the summary work using CAD?