In this dissertation, we extend database models and query languages to support spatio-temporal information, including representations for changing positions and shapes. Furthermore, we propose techniques to support spatio-temporal extensions on Object-Relational systems and achieve end-user extensibility.; We begin from temporal data models and query languages. Our approach is based upon a point-based representation of time enhanced with user-defined aggregates that support interval-oriented operators such as DURATION and DURING. This approach provides several advantages, since it solves the coalescing problem, minimizes extensions required from SQL, and it is applicable to all query languages.; Then, we extend the time model at the physical level to spatio-temporal databases. As the basis of our implementation, we introduce counterclockwise-directed are introduced to support spatial operators such as CONTAIN and OVERLAP, and spatial-temporal operators such as MOVING_DISTANCE. Finally, we represent moving objects as sequences of snapshots of spatial objects such as points, lines and polygons.; To reconcile logical and physical requirements, we follow an implementation approach based on a layered architecture. Thus, for temporal information, point-based representations are mapped to an interval-oriented representation at the internal level. Likewise, we map polygon-based representations into a triangle-based internal representation. We implement these mappings through query transformations, and user-defined aggregates.; A final advantage of our approach is that it allows end-users to further extend and customize their system: in fact in our SQLST system, users can extend database functionality by writing new user-defined aggregates in SQLST itself.; In summary, the spatio-temporal data models and query languages introduced in this dissertation offer several conceptual advantages; furthermore, our implementation approach provides considerable practical benefits, including compatibility with Object-Relational systems, flexibility, robustness, and ease of customization by end-users.
展开▼
机译：在本文中，我们扩展了数据库模型和查询语言以支持时空信息，包括用于改变位置和形状的表示。此外，我们提出了在对象关系系统上支持时空扩展并实现最终用户可扩展性的技术。我们从时态数据模型和查询语言开始。我们的方法基于时间的基于点的表示，该时间通过用户定义的聚合来增强，这些聚合支持面向间隔的运算符，例如DURATION和DURING。这种方法具有多个优点，因为它解决了合并问题，使SQL所需的扩展最小化，并且适用于所有查询语言。然后，我们将时间模型在物理级别扩展到时空数据库。作为实现的基础，我们引入了逆时针定向以支持诸如CONTAIN和OVERLAP之类的空间算子以及诸如MOVING_DISTANCE之类的时空算子。最后，我们将运动对象表示为空间对象（如点，线和多边形）的快照序列。为了调和逻辑和物理要求，我们遵循基于分层体系结构的实现方法。因此，对于时间信息，基于点的表示在内部级别映射为面向间隔的表示。同样，我们将基于多边形的表示形式映射为基于三角形的内部表示形式。我们通过查询转换和用户定义的聚合来实现这些映射。我们方法的最后一个优点是，它允许最终用户进一步扩展和定制他们的系统：实际上，在我们的SQL ST 系统中，用户可以通过在SQL中编写新的用户定义的聚合来扩展数据库功能。 ST 本身。综上所述，本文介绍的时空数据模型和查询语言具有一些概念上的优势。此外，我们的实现方法还具有可观的实际好处，包括与对象关系系统的兼容性，灵活性，健壮性以及最终用户自定义的简便性。
展开▼
