Data management and data mining in spatio-temporal databases

时空数据正在以越来越快的速度收集，例如通过卫星图像、GPS系统、移动通信网络、医学成像技术、传感器网络或视频观测等。在本提案中，我们将“原始”形式的时空数据视为（部分）由位置和/或物体在空间参照系中随时间的扩展所描述的数据。空间参照系可以是，例如，地球表面的一部分，城市的街道网络，人类大脑的模型，建筑物，由远程操作的车辆探索的水下区域的一部分，等等。时空数据通常表示移动的物体（例如，带着启用GPS的设备四处移动的人）或移动和进化的几何形状（例如，森林火灾）。这类数据可用性的增加为移动计算和商业、车队控制、环境和社会经济现象的监测和分析、交通控制、动物跟踪、肿瘤放射治疗计划、流行病学分析等应用领域的高级服务和数据分析提供了巨大的机会。然而，针对大容量时空数据的先进数据分析和查询技术仍处于起步阶段。一方面，地理信息系统（GIS）虽然提供了复杂的空间数据分析和显示方法，但还没有准备好处理大数据量和探索性时空分析的复杂性。另一方面，对象关系数据库系统为非空间数据和一定程度上的空间数据提供了有效的访问方法，但还不能有效和高效地支持时空数据（尽管近年来在轨迹数据建模和索引用于基本查询处理方面的研究取得了显著进展）。本研究计划的全球目标是弥合这一差距，并设计通用数据库技术，通过开发有效的方法和算法来表示、存储、查询和挖掘不同尺度的时空数据，以支持广泛的高级应用、分析和服务。

Spatio-temporal data is being collected at an increasing rate, e.g., through satellite images, GPS systems, mobile communication networks, medical imaging technology, sensor networks, or video observations etc. In this proposal, we consider spatio-temporal data in its "raw" form to be data that is (partially) described by locations and/or extensions of objects in a spatial frame of reference over time. Spatial frames of reference can be, for instance, parts of the earth's surface, a city's street network, a model of the brain of humans, a building, parts of an underwater area explored by a remotely operated vehicle, etc. Spatio-temporal data often represents moving objects (e.g., people moving around with a GPS enabled device) or moving and evolving geometries (e.g., forest fires). The increased availability of this type of data presents a great opportunity for advanced services and data analyses in application areas such as mobile computing and commerce, fleet control, monitoring and analyzing of environmental and socioeconomic phenomena, traffic control, animal tracking, radiation treatment planning for tumours, epidemiological analyses, etc. However, advanced data analysis and querying techniques for large volumes of spatio-temporal data are still in its early stages. On the one hand, geographic information systems (GIS), although they provide sophisticated analysis and display methods for spatial data, are not yet ready to deal with large data volumes and the complexity of exploratory spatio-temporal analysis. On the other hand, object-relational database systems, which provide efficient access methods for non-spatial and to some degree spatial data, do not yet support spatio-temporal data in an effective and efficient way (although research in recent years has advanced significantly in the area of modeling and indexing trajectory data for basic query processing). The global objective of this research program is to bridge this gap, and to devise general database technology to support a wide range of advanced applications, analyses, and services by developing effective and efficient methods and algorithms for representing, storing, querying, and mining of spatio-temporal data at different scales.