Improvement of task-oriented visual interpretation of VGI point data (TOVIP)

志愿地理信息（VGI）已经在各种各样的社会和商业应用中显示出巨大的潜力。通过各种服务和软件解决方案，专家和非专业人员都能够通过互联网收集和显示数据。VGI通常作为点数据生成，表示兴趣点、其他质量或数量。典型的例子是环境数据（如交通噪音、PM10值、交通量的测量值），或有关事故或犯罪地点的数据。VGI数据通常显示非常大的数据量以及语义和时间的异质性。这两个方面都可能极大地降低视觉表示和探索的可用性，特别是在对高级（概要）模式的解释感兴趣的情况下。如果重点放在点数据上，则渲染性能下降或几何或主题点杂波的影响是可能的。通常，应用泛化方法来克服这些杂波问题。这个项目的重点将是优化为特定的可视化解释任务（比如检测热点或极值）而设计的泛化工作流，而不是只关注整体方法或孤立的泛化操作。在使用基于约束的方法时，需要考虑两个潜在的相互矛盾的方面：空间模式的保存和可读性。然而，基于约束的方法在定义约束方面仍然存在局限性。此外，通过约束触发泛化过程的研究目前还很有限。特别是VGI点数据要么是在多个尺度水平上产生的，要么是在更长的时间内产生的，甚至是实时的，这两者都需要非静态显示。然而，这种表示的泛化过程的改进尚未得到彻底的研究。例如，交互式多尺度视图需要考虑尺度转换，也就是说，对约束的更改必须形式化为正在考虑的任务的函数。当使用静态或甚至移动点的多时间表示时，还必须考虑与这些动画相关的额外复杂性和认知工作量的限制。因此，该项目的总体目标是提高VGI点数据显示的视觉可解释性——考虑到基于静态、多尺度或多时间显示的特定高级（概要）任务。从方法论的角度来看，项目从相关概要任务的定义开始。分析性和经验性调查定义了最少的约束。在此基础上，开发了基于智能体的模型，以优化整个泛化过程。最后，实证研究评估了扩展约束集的假设进展以及提出的基于智能体的优化方法。

Volunteered Geographic Information (VGI) has already shown great potential for a huge variety of social and commercial applications. With a variety of services and software solutions, both experts and non-expert are able to collect and to display data via the internet.VGI is very often generated as point data, representing points of interest, other qualities or quantities. Typical examples are environmental data (such as measurements of traffic noise, PM10 values, traffic volume), or data about accident or crime spots.VGI data typically shows a very large volume of data as well as semantic and temporal heterogeneity. Both aspects can drastically reduce the usability in visual presentation and exploration, in particular, if the interpretation of high-level (synoptic) patterns is of interest. If the focus is on point data, a decline in rendering performance or the effects of geometric or thematic point clutter are possible.Typically, generalization methods are applied in order to overcome these clutter problems. Instead of looking at a holistic approach or isolated generalization operations only, the focus in this project will be on optimizing generalization workflows designed for specific visual interpretation tasks (such as detecting hot spots or extreme values). When using constraint-based approaches, there are two potentially contradictory aspects to consider: preservation and readability of the spatial patterns. However, constraint-based approaches still have limitations in defining constraints. In addition, research to trigger the generalization process through constraints has been quite limited so far.Especially VGI point data is produced either in multiple scale levels, or over longer periods of time or even in real-time, both requiring non-static displays. However, the improvement in the generalization process of such representations has not been thoroughly investigated. As an example, interactive multi-scale views require consideration of scale transitions, i.e., changes to constraints must be formalized as a function of the task being considered. When multi-temporal representations of static or even moving points are used, the added complexity and limitations of the cognitive workload related to these animations must also be considered.Consequently, the overall goal of this project is to improve the visual interpretability of VGI point data displays – taking into account specific high-level (synoptic) tasks based on static, multi-scale or multi-temporal displays.From a methodological point of view, the project begins with the definition of relevant synoptic tasks. Analytical and empirical investigations define a minimum of constraints. Based on this, agent-based models are developed with the aim of optimizing the entire generalization process. Finally, empirical studies evaluate the assumed progress of the extended set of constraints together with the proposed agent-based optimization method.