ABSTRACT

Assessing OpenStreetMap (OSM) data quality against authoritative data sources may not always be viable. This is primarily because of the multi-dimensional nature and heterogeneity of the maps, yet the activity is pivotal for targeted data cleansing and quality enhancement undertakings in these data sets. A salient facet of OSM, allowing contributors to flag potential problems encountered during the mapping process, is the FIXME tag. In this article, we examine and discuss OSM data quality through the vast expanse of issues (knowledge) documented via FIXME. We present a classification and analysis of these quality issues, exposed as topic models and grounded in the ISO-19157 standard, across USA and Australia. Regional distributions of these topics are further qualitatively analyzed to ascertain the variation of key issues in OSM. We also present a comparison of the intrinsic issue classification against those identified in an issue corpus of an authoritative map data source. Due to the considerable heterogeneity in user mapping and reporting, OSM issue detection and classification remains problematic. This research presents a flexible and intrinsic data-mining approach, linking established ISO data quality standards to OSM issue categorization. Our work, thus informs the development of automated error correction methods for VGI datasets.

摘要

根据权威数据源评估 OpenStreetMap (OSM) 数据质量可能并不总是可行的。这主要是因为地图的多维性质和异质性，但该活动对于这些数据集中的有针对性的数据清理和质量提升工作至关重要。OSM 的一个显着方面，允许贡献者标记潜力映射过程中遇到的问题，就是FIXME标签。在本文中，我们通过 FIXME 记录的大量问题（知识）来检查和讨论 OSM 数据质量。我们对这些质量问题进行了分类和分析，这些问题以主题模型的形式公开，并以 ISO-19157 标准为基础，遍及美国和澳大利亚。进一步定性分析这些主题的区域分布，以确定 OSM 中关键问题的变化。我们还将内在问题分类与权威地图数据源的问题语料库中识别的问题分类进行了比较。由于用户映射和报告存在相当大的异质性，OSM 问题检测和分类仍然存在问题。这项研究提出了一种灵活的内在数据挖掘方法，将已建立的 ISO 数据质量标准与 OSM 问题分类联系起来。因此，我们的工作为 VGI 数据集的自动纠错方法的开发提供了信息。