Given a file with parse trees, implement code that does the following for each parse tree:
1. Loops through all the leaves and selects the verbs.
2. For each verb, selects potential arguments following the heuristic we saw in class (Xue and Palmer)
3. For each verb and potential argument, extracts at least the following features:
• the verb form (the actual predicate),
• the POS of the verb,
• the number of leaves in the potential argument, and whether the argument occurs before or after the verb (i.e., the direction).