Keeping Interval-Based Functional Dependencies Up-to-Date

In the temporal database literature, every fact stored in a database may be equipped with two temporal dimensions: the valid time, which describes the time when the fact is true in the modeled reality, and the transaction time, which describes the time when the fact is current in the database and can be retrieved. Temporal functional dependencies (TFDs) add valid time to classical functional dependencies (FDs) in order to express database integrity constraints over the flow of time. Currently, proposals dealing with TFDs adopt a point-based approach, where tuples hold at specific time points, to express integrity constraints such as “for each month, the salary of an employee depends only on his role”. To the best of our knowledge, there are no proposals dealing with interval-based temporal functional dependencies (ITFDs), where the associated valid time is represented by an interval and there is the need of representing both point-based and interval-based data dependencies. In this paper, we propose ITFDs based on Allen’s interval relations and discuss their expressive power with respect to other TFDs proposed in the literature: ITFDs allow us to express interval-based data dependencies, which cannot be expressed through the existing point-based TFDs. ITFDs allow one to express constraints such as “employees starting to work the same day with the same role get the same salary” or “employees with a given role working on a project cannot start to work with the same role on another project that will end before the first one”. Furthermore, we propose new algorithms based on B-trees to efficiently verify the satisfaction of ITFDs in a temporal database. These algorithms guarantee that, starting from a relation satisfying a set of ITFDs, the updated relation still satisfies the given ITFDs. 1 An example of interval-based constraints Most health care institutions collect a large quantity of clinical information about patient and physician actions, such as therapies and surgeries, as well as about health care processes, such as admissions, discharges, and exam requests. All these pieces of information are temporal in nature and the associated temporal dimension needs to be carefully considered in order to be able to properly represent clinical data and to reason about them [2]. In this section, we briefly ⋆ A short summary of the results published in [3] and [4]. # TherType PatId Phys DrugCode Qty B E 1 antiviral 1 Dorian 0458 300 1 16 2 analgesics 1 Cox 0976 200 2 10 3 cardiovascular 1 Turk 0118 100 3 8 4 antipyretics 1 Cox 0976 100 9 11 5 sedative 1 Turk 0345 10 13 15 6 anxiolytic 1 Dorian 0345 10 17 19 7 antiviral 2 Kelso 0458 200 1 10 8 cardiovascular 2 Quinlan 0118 100 4 7 9 analgesics 2 Reid 0976 150 5 9 10 antiviral 2 Reid 0458 300 8 14 11 antiviral 1 Dorian 0789 200 1 18