vgiReports: Accessible Reporting of Spatiotemporal Geographic Information Leveraging Generated Text and Visualization

遵循公益事业也应该为公众服务的理念，我们的目标是让志愿者提供的地理数据和衍生的分析结果可供广泛的受众使用。由于数据描述需要特别的自解释性以及广泛适用性的表达性，因此文本描述和可视化将相互补充。文本提供了自包含的解释，而可视化则添加了用户可以快速感知和处理的信息丰富的表示。然而，地理数据往往是复杂的，特别是当它同时包含空间和时间成分时。因此，对数据进行分析和表示的挑战是，例如，识别信息和结果的正确子集，集成不同级别的数据抽象，以及使底层算法数据分析透明和可信。以前对地理数据的研究，包括各种视觉分析系统，主要集中在专业和专家用户。相比之下，本项目的目标是研究时空地理信息的可理解和自解释表示。我们计划研究表示数据的基本方法，以及研究与应用和用户接受相关的问题。专家撰写的样本报告将构成研究人类如何抽象、描述和说明复杂时空模式的基础，以获得可访问的报告。结合自然语言生成和地理可视化技术，我们将按照这些示例自动创建报告。交互式改进将允许用户根据他们的偏好对报告进行个性化设置，并从他们的角度探索数据。不同类型的时空数据——例如，地理参考文件、运动轨迹或网络——需要开发新的总结和报告技术。基于这些技术，我们的方法适用于各种应用领域，包括城市生活、交通、环境场景和紧急情况。我们计划与来自各个领域的专家一起为其中几个应用领域构建示例系统。在用户研究中对这些示范进行评估，将有助于了解公众和决策者接受这种数据丰富的交互式报告的标准是什么。

Following the idea that a public good should also serve the public, it is our goal to make volunteered geographic data and derived analysis results accessible for a wide audience. Since data descriptions need to be particularly self-explaining as well as expressive for a wide applicability, textual description and visualizations will complement each other. Text provides self-contained explanations while visualization adds information-rich representations that can be perceived and processed quickly by users. However, geographic data is often complex especially if it has both a spatial as well as a temporal component. Resulting challenges for analysis and presentation of the data are, for instance, to identify the right subset of information and results, to integrate various levels of data abstraction, and to make the underlying algorithmic data analysis transparent and trustable. Previous research on geographic data, including various visual analytics systems, mostly focused on professional and expert users. In contrast, the goal of the proposed project is to research understandable and self-explaining representations of spatiotemporal geographic information. We plan to investigate fundamental methods to represent the data as well as research questions related to application and user acceptance. Expert-authored sample reports will form a basis to study how humans abstract, describe, and, illustrate complex spatiotemporal patterns for accessible reporting. Combining natural language generation and geographic visualization techniques, we will automatically create reports following these examples. Interactive refinement will allow users to personalize the reports according to their preferences and to explore the data from their perspective. Different types of spatiotemporal data – for instance, geo-referenced documents, movement trajectories, or networks – require the development of new summarization and reporting techniques. Based on these techniques, our approach is applicable to various application areas including city life, transportation, environmental scenarios, and emergency situations. We plan to build example systems for several of these application areas together with experts from the respective fields. Evaluating these demonstrators in user studies will provide insights what are the criteria for acceptance of such data-rich interactive reports by users from the general public and decision makers.