HeiGIT and GIScience Contributions to Last Year’s Conference-athon

Featured Figure: Object history revisions from Alishiba Dsouza of the University of Bonn, GIScience’s Moritz Schott and Prof. Dr. Sven Lautenbach delivered the academic track talk, “Comparative Integration Potential Analyses of OSM and Wikidata – the Case Study of Railway Stations” slides.

With a new year of papers, workshops, and publications ahead of us, the team at GISciene and HeiGIT are looking back and building on the many contributions to last year’s conferences. During the previous summer’s FOSS-Community “conference-athon” in Florence, when QGIS-Meeting, State of the Map, HOT unSummit and FOSS4G all occurred in the same short span, the team participated in a flurry of events showcasing ohsome, case studies, and community projects.

Beginning with State of the Map 2022, HeiGIT’s Benjamin Herfort, Michael Auer, Matthias Schaub, and Rafael Troilo of the Big Spatial Data Analytics team led the Saturday workshop “Analyzing changes in OSM over time – full history access to OSM data through the ohsome framework” building upon the OpenStreetMap History Database (OSHDB) and ohsome framework, which were introduced to the OSM community at State of the Map 2018 in Milan. The idea for OSHDB itself is over a decade old, stemming from HeiGIT’s Scientific and Managing Director Prof. Dr. Alexander Zipf in 2010. OSHDB Version 1.0 was released this past December, with a press release, a review of the development process, and a social media campaign of favorite use cases and reflections to celebrate the announcement. Benjamin Herfort and Rafael Troilo outlined OSHDB and ohsome for workshop participants and then walked them through hands-on examples focusing on re-tagging, OSM-based metrics, and contributions in the OSM space.

The next day, Alishiba Dsouza of the University of Bonn, GIScience’s Moritz Schott and Prof. Dr. Sven Lautenbach delivered the academic track talk, “Comparative Integration Potential Analyses of OSM and Wikidata – the Case Study of Railway Stations” based on their short paper of the same name. The presentation, with slides and video available online, analyzes a series of comparative data insights to explore the potential and implications of integration between knowledge graphs and OSM using the ohsome api for their dataset and community metrics. The researchers compare OSM to Wikidata in size, structure, and distribution and then discuss the respective online communities for each tool. Using railway stations as a case study, the authors found divergent trends in the coverage and amount of contributions in the two communities, elucidating differences in the goals and focuses of each group.

Following this, Benjamin Herfort led the 5-minute academic track lightning talk, “Inequalities in the completeness of OpenStreetMap buildings in urban centers” drawing from his short paper. The talk, with slides and video available, discusses the stark data inequalities in OSM data. These inequalities have persisted despite humanitarian- and corporate-organized mapping activities, as shown by an analysis using a machine learning regression method based on generalized additive models (GAMs) to calculate the OSM building completeness for 13,135 urban centers. The results point to the need to apply multi-scale building completeness measures to account for missing data and the necessity of completeness maps and socio-demographic information to make sure “nobody is left behind.”

Figure: From the slides of Benjamin Herfort’s 5-minute academic track lightning talk, “Inequalities in the completeness of OpenStreetMap buildings in urban centers” showing completeness of OSM building stock in urban areas.5 minutes was also just enough time for GIScience’s Moritz Schott, Prof. Dr. Sven Lautenbach, and Prof. Dr. Alexander Zipf as well as Technische Universität Berlin’s Adina Zell and Prof. Dr. Begüm Demir to present the talk, “Returning the favor – Leveraging quality insights of OpenStreetMap-based land-use/land-cover multi-label modeling to the community.” The group endeavored to prove the fitness of OSM for multi-label classification and presented a workflow to enhance OSM-based multi-labels using machine learning, with results available via the HOT Tasking Manager. This proposed tool provides an automated OSM multi-label extraction, modeling and verification procedure including a return of results to the OSM community for feedback and quality assertion. This approach allows for large scale data assessments and repeated OSM LULC quality and completeness estimations for large areas over time. In addition, local OSM data can be used for modeling leading to regional homogenization and data cohesion within regional communities. As always the respective code is available under an open licence and you can try out the developed custom sandbox-Tasking Manager.

In addition to the workshops and presentations, Marcel Reimuth, Maximiliane Kitzunger, and Prof. Dr. Alexander Zipf also participated in the call-for-posters with thier contribution on access to safe abortions in Germany, detailing current accessibility by car, the situation for abortion seekers who may require a surgical abortion or may not speak German, and how the situation might evolve after the repeal of a law banning the advertisement of abortion.

Figure: Visual from former HeiGIT colleague Munazza Usmani’s poster highlighting a case study of Beirut where change detection analysis was performed using remote sensing and OpenStreetMap.Munazza Usmani, a former HeiGIT colleague completing her PhD at the University of Trento, presented her poster highlighting a case study of Beirut where change detection analysis was performed using remote sensing and OpenStreetMap. Munazza and her colleagues developed a self-supervised, systematic method for disaster mapping using a CNN-based model for building segmentation and change detection and OSM data to train their deep learning network.

Finally, Moritz Schott held a talk at FOSS4G presenting OSM Element Vectorization, a new tool that enables users to calculate a large set of quality indicators for single OSM elements. By integrating multiple aspects of the OSM ecosystem into one workflow that allows the quantitative assessment of selected OSM elements or all elements in a defined region, a vectorization of single OSM elements is produced as a machine-readable result. In an example study, the tool was used to uncover some interesting tendencies in OSM’s land-use data.

Last year’s conference-athon gave HeiGIT and GIScience researches ample opportunity to present the most cutting-edge research in the geospatial technology and geography space. As submissions begin for this summer’s conferences, we look forward to expanding upon our past work and introducing the world to the next round of projects to enable faster routing, quality research, and geospatial analysis. To stay updated on all the promising research to come, check out our GIScience and HeiGIT blogs and make sure to follow us on Twitter, LinkedIn, and Mapstodon.