Previous: The Proposed Query System Up: A system for Querying the Text Track in Movies Next: Future Extensions

Implementation Stages

1. Creation of a DTD for the Movie Script Structure: A DTD [6][2] will first be created using some of the examples that we have presently available. This DTD might not suffice for every different kind of movie, so care needs to be taken to make the design flexible enough for later extension of the DTD.
2. Hand Creation of a few movies in SGML format: Since we do not have data for testing the code, a few movies will be semi-automatically coded into SGML using the DTD designed in the previous step.
3. Deciding on a Data Structure for internal storage of the data: The next, and possibly the most important stage in the project, will be deciding on (or designing) a data structure to implement the complex track. The data structure should be able to handle indexing on the various regions/tags, connecting to the frames using the Exodus/Shore storage IDs,[1][7] and to incorporate equivalent operations like ``shift'', ``expand'', ``contract'', etc., on the text track.
4. Using an SGML Parser to parse the SGML scripts into the data structure, and link it up with the frame OIDs. The input scripts will then be parsed using a standard SGML parser (e.g. sgmls or PAT[4]), and inserted into the data structure implemented in a previous step.
5. Designing a simple query language and interface for information retrieval from the script database: A simple query language with the ability to query on various aspects of the script, with the end result as either part of the script itself, or part of the movie, will be designed, and an interface for this query language would be implemented. The exact nature of this interface is still an undecided issue, but it might be similar to the QBE interface in relational databases, and would be implemented in TCL/TK.[5]