Introduction

Currently, the track data model treats every object as a flat series of atoms [3]. It is not clear how one is to model objects that have a more complex structure than is indicated by such a flat list. For example, a movie script is more than merely an ordered collection of lines.

Acts, along with attributes such as titles, authors, etc., compose a script. Scenes compose acts. Lines compose scenes. Perhaps one could devise a scheme that shows these interrelationships using a multitude of tracks, e.g., one track for acts, one track for scenes, etc. However, this seems extremely fragmented. Also, it remains unclear where the model would store metadata such as the script's title that do not really have any place in a linear order.

Another approach is possible. One can create a complex data structure to model the intricacies of a script while still using tracks to model the movie to which the script belongs. At an appropriate point of abstraction in the script, say with the lines of the script, one could interconnect the track atoms and the script components. Each line could have pointers, say, to the first and last frames in which a character speaks the line. This method preserves all the benefits of the track model for the movie's frames and sounds while allowing one to model all the additional structure of a movie script.

asengupt@center