Search

There have been significant research activities in Temporal Databases during the last decade. However, the developments of a semantics of time, a temporal model for efficient database systems and temporal query languages still need much study. Based on the researches of the TDB group [Snodgrass 1987], the review of research about TDBMS in this dissertation mainly emphasises three aspects as follows. 1) The formulation of a semantics of time at the conceptual level. A topology of time and types of time attributes are introduced. A new taxonomy for time attributes is presented: assertion time, event time, and recording time. 2) The development of a model for TDBMS analogous to relational databases. Based on Snodgrass' classification, four kinds of databases: snapshot, rollback, historical and temporal are discussed in depth. But the discussion distinguishes some important differences from the representation of the TDB model: - historical relation for most enterprises is an interval relation, but not a sequence of snapshot slices indexed by valid time. The term "tuple" no longer simply refers to an entity as in traditional relational databases. It refers to different level representations of an object: entity, entity state, observation of entity, and observation of entity state in different types of databases. 3) The design of temporal query languages. We do not present a new temporal query language in this dissertation, but we discuss a Quel-like temporal query language, TQuel, in some depth. TQuel is compared with two other temporal query languages TOSQL and Legol 2.0. We centre the main discussion on TQuel's semantics for tuple calculus. The classification for the relationships between overlapping intervals suggests an approach using temporal logic to classify the derived tuples in tuple calculus. Under such an approach, a new presentation for tuple modification calculus is proposed, not only for interval relations, but also for event relations.