Retrieving transportation information with Constraint SQL and Constraint Datalog.

摘要

If transportation planners, operators, designers, policy makers, and other users of transportation information have to recreate how they retrieve their information when managers of the information make changes, then the managers must choose between burdening the users with the change and burdening themselves with the current state. This is a problem because there are costs associated with both choices: the cost of having the users recreate how they retrieve their information and the cost of not making the change. This dissertation identifies (1) the domain ordering problem, (2) the retrieval optimization problem, (3) the changing data models problem, (4) the lossy data compression problem, and (5) the non-standard units problem. It demonstrates how these five problems allow a manager's changes to affect how users retrieve their information. It develops an approach in which users can retrieve information so that managers' changes do not affect how they retrieved their information. It then applies that approach to Constraint SQL and Constraint Datalog, adding the necessary features for both to completely address all five problems. These additional features include unit handling, significant figures, a "selected" relation, the extended use of views, and, in Constraint Datalog, the ASC and DESC functions, the curly brackets, and domain reordering. Other features include reserved attribute names, case sensitivity, and, in Constraint Datalog, the asterisk operator, the use of blanks, and stopping short. It also introduces the "creator ID" to allow users to share user-created functions and views, and it introduces a standards committee to adopt standard functions and views. Ultimately, this dissertation hypothesizes that, by addressing all five problems, data languages like Constraint SQL and Constraint Datalog will allow managers to implement changes without affecting users' queries, thereby allowing managers to freely adopt the newest optimization techniques, the newest data models, the newest data lifecycle management techniques, and the most appropriate units.