AitF: Collaborative Research: Modeling movement on transportation networks using uncertain data

在当前以数据为中心的时代，有许多高度多样化的数据源提供有关交通网络上移动的信息。示例包括GPS轨迹、社交媒体数据和交通流量测量。由于不频繁的采样和稀疏的覆盖范围所导致的固有的不确定性，这些运动数据中的大部分是具有挑战性的。该项目的目标是开发一个统一的框架，使用尽可能多的可用数据源，从数据中自动提取有意义的交通和移动信息。将开发概率网络移动模型，以捕获网络上随时间推移的移动概率和流量。研究结果将影响一系列依赖于捕捉人口流动的应用，如城市规划、地理营销、交通管理和应急管理。教育活动将贯穿整个项目。学生将密切参与研究和实际实施，并将接受时空数据管理，算法开发和（轨迹）数据分析方面的培训。这些技能的结合在空间数据科学中越来越重要。该项目涉及的专题将丰富这两个机构的课程材料和课程编制。 该项目的目标是创建一个统一的框架，用于聚合和分析道路网络上各种不确定的运动数据，旨在提供查询和预测交通量和运动的工具。将开发网络上的概率运动模型，可以处理异构数据源，包括GPS轨迹，地理标记的社交媒体数据，自行车共享数据，公共交通数据和交通量数据。这些数据的多样性和时空不确定性将通过贝叶斯交通模式学习方法来解决，该方法首先使用更确定的数据训练运动模型，然后将其用于填补更不确定数据中的空白。该项目将推进理论界（计算几何学，数据挖掘）以及应用界（空间数据库，位置科学）的最新技术。研究结果将在项目网站（movementanalytics.org）上公布，并将通过在会议上的介绍和演示以及通过在期刊上的出版物在著名场所传播。

In the current data-centered era, there are many highly diverse data sources that provide information about movement on transportation networks. Examples include GPS trajectories, social media data, and traffic flow measurements. Much of this movement data is challenging to utilize due to the inherent uncertainty caused by infrequent sampling and sparse coverage. The goal of this project is to develop a unified framework that uses as many available data sources as possible to extract meaningful traffic and movement information automatically from the data. Probabilistic network movement models will be developed that capture movement probabilities and traffic volume on a network over time. The results will impact a range of applications that rely on capturing population movements, such as urban planning, geomarketing, traffic management, and emergency management. Educational activities will be integrated throughout the project. Students will be closely involved in research and practical implementations, and will be trained in spatio-temporal data management, algorithms development, and (trajectory) data analysis. The combination of such skills is increasingly important in spatial data science. Topics involved in this project will enrich the course material and curriculum development at both institutions. The objective of this project is to create a unified framework for aggregating and analyzing diverse and uncertain movement data on road networks, with the aim to provide tools for querying and predicting traffic volume and movement. Probabilistic movement models on the network will be developed that can handle heterogeneous data sources, including GPS trajectories, geo-tagged social media data, bike-share data, public transport data, and traffic volume data. The diversity and spatio-temporal uncertainty of this data will be addressed with a Bayesian traffic pattern learning approach that first trains the movement models with the more certain data, which in turn will be used to fill gaps in the more uncertain data. The project will advance the state-of-the-art in theoretical communities (computational geometry, data mining) as well as in applied communities (spatial databases, location science). The results of the research will available on the project website (movementanalytics.org), and will be disseminated in prestigious venues through presentations and demonstrations at conferences, and through publications in journals.