Questions about indices

[d223302]

Hi, thank you for releasing such useful resources.
I have some questions about the .npz files released.
In the Readme, it reads

Each file can be loaded with numpy.load and is a dictionary mapping from DBPedia URI strings to numpy arrays of pre-training dataset indices where that entity occurs.

I am not very sure what the indices here refer to. To put it another way, how can I obtain the text corresponds to an index in the pre-training dataset?

[zhentingqi]

Same question here. Is there any update?