Optimized spatial data conflation with topological conditions

本项目致力于研究地理数据合并，即合并同一地理区域的两个独立数据集的过程。特别是，该项目的目标是开发能够自动合并过程的算法。传统上，合并是通过耗时的手动过程实现的，这可能不可靠，实现起来也很昂贵。通过开发可融入地理信息系统的方法，该项目增加了不同组织以可靠和可重复的方式进行更大规模合并的潜力。该项目还有助于培训和教育多个研究生。该项目的总体目标是改进地图对象匹配和合并的计算方法，比受过培训的人类从业人员更可靠和更好地进行匹配和合并。为了实现这一目标，研究人员开发了自动化方法，这些方法基于据称是人类使用的逻辑背后的策略。第一种策略是误差最小化，即人类倾向于根据空间对象的相似性或距离来匹配它们。这被称为优化合并，其中对象匹配被视为最小化差异的自然问题。第二种策略是使用空间上下文来处理更复杂、更模糊的匹配情况。后一种观点在以前的实现中基本上没有。通过允许计算机在相邻对象中合并相同的上下文条件，新模型将生成空间上一致的匹配，从而允许人类从业者专注于解决一小部分异常情况。由于可靠性的提高，该项目的新模型和算法可以减少合并的金钱和时间成本，使其成为学生、规划者、地理学和其他空间学科的研究人员以及地理信息系统(GIS)的普通用户都可以使用的工具。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This project addresses the study of geographic data conflation, which is the process of merging two separate datasets of the same geographical region. In particular, the goal of the project is to develop algorithms that can automate the process of conflation. Conventionally, conflation has been achieved by time-consuming manual processes, which can be unreliable and expensive to implement. By developing methods that can be integrated into geographic information systems, this project increases the potential for diverse organizations to conduct conflation at larger scales in reliable and reproducible ways. The project also contributes to the training and education of multiple graduate students.The overarching objective of this project is to advance computational methods for matching and merging map objects reliably and better than trained human practitioners. To accomplish this goal, the researchers developed automated methods that are based on strategies purportedly underlying the logic used by humans. The first strategy is error minimization, in which humans tend to match individual spatial objects based on their similarity or distance. This is known as optimized conflation, in which object matching is treated as a natural problem of minimizing discrepancy. The second strategy is to use spatial context to deal with the more complex, ambiguous matching cases. This latter perspective has been largely absent from previous implementations. By allowing computers to incorporate the same contextual conditions in neighboring objects, the new models will generate spatially consistent matches, thus allowing human practitioners to focus on resolving a small percentage of anomalous cases. Due to the increased reliability, the new models and algorithms of this project can reduce the monetary and time cost of conflation, making it a tool that can be used by students, planners, researchers in geography and other spatial disciplines as well as the average user of Geographic Information Systems (GIS).This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.