Action Mapping of "select" operation in Mind2web

Dear team,

I noted that you have released training examples of Mind2web, where the "select" operation (select a value from a dropdown button) is represented as "select(start_box='<|box_start|>(x1,y1)<|box_end|>')", so does the (x1, y1) here means the center point of the expanded bbox to be selected? Or, the (x1, y1) means the center point of the dropdown button, but in this way, the value to be selected is missing.

Thanks you!

Read the released Mind2web training examples and inspect how the select operation is represented. Clarify whether the coordinates refer to the expanded option or the dropdown button, and explain how the value to select is identified; the issue is complete when this ambiguity is documented.

Written by the indexing model from the issue text.