Abstract |
A database contains data about the reality that it models. This data must be precise, complete and consistent so that various applications can use it. Traditional databases contain only a snapshot of the modeled reality which includes the more recent data. As the database evolves, new data values replace old ones. However, the increasing need to retain the history of the database contents proves traditional databases inefficinet to cover the needs of significant number of applications. Therefore, there is recently an increasing interest in studying and extending existing data models, with a temporal dimension. This work is a study on incorporateing a time dimension into the SIS_telos structural, object-oriented data modle in order to model mutable object behavior. This temporal extension can lead the database to inconsistent states. One of the goals of this work is to study and define the necessary constraints that ensure the consistency of the extended temporal data model. Additionally, we study the potential of managing such a time-varying data behavior. Thus, we define a temporal algebra that includes operators for retrieving temporal extended objects and for expressing queries concerning the temporal dimension of the data model. This algebra manipulates simple data objects as well as paths. Paths are handled as if they were objects themselves. We define the temporal dimension of a path as the intersection of the temporal dimensions of the links participating in this path. A path ``exists`` temporally, only if the above intersection is non-empty. To compute temporally existing paths, an algorithm is defined and its complexity is studied. We demonstrate that the temporal dimnension can easily be modeled using the existing knowledge representation of SIS-Telos. Such an extension can be supported without any major changes in the philosophy of SIS.
|