Each amendment essentially describes a change to the original text. For the next problems, it is required for the system to understand the nature of the change. Essentially, it needs a model of each change.
The model requires:
Location information; Where does the change need to be made in the original text.
Action information; How does the original text need to be modified.
Operand information; Depending on the action information.
This is essentially an information extraction problem.