Key points are not available for this paper at this time.
The data extraction systems studied in the MUC-3 evaluation perform a variety of subtasks in filling out templates. Some of these tasks are quite complex, and seem to require a system to represent the structure of a text in some detail to perform the task successfully. Capturing reference relations between slot fillers, distinguishing between historic and recent events, and many other subtasks appear to have this character.
David Lewis (Tue,) studied this question.