Centaur VGI: A Hybrid Human–Machine Approach to Address Global Inequalities in Map Coverage

Abstract

Despite advances in mapping technologies and spatial data capabilities, global mapping inequalities are not declining. Inequalities in the coverage, quality, and currency of mapping persist, with significant gaps in remote and rural parts of the Global South. These regions, representing some of the most economically and resource-disadvantaged societies in the world, need high-quality mapping to aid in the delivery of essential services, such as health care, in response to severe challenges such as poverty, conflict, and global climate change. Volunteered geographic information (VGI) has shown potential as a solution to mapping inequalities. Contributions have largely been made in urban areas or in response to acute emergencies (e.g., earthquakes or floods), however, leaving rural regions that suffer from chronic humanitarian crises undermapped. An alternative solution is needed that harnesses the power of volunteer mapping more effectively to address regions in most need. Machine learning holds promise. In this article we propose centaur VGI, a hybrid system that combines the spatial cognitive abilities of human volunteers with the speed and efficiency of a machine. We argue that centaur VGI can contribute to mitigating some of the political and technological factors that produce inequalities in VGI mapping coverage and do so in the context of a case study in Acholi, northern Uganda, an inadequately mapped region in which the authors have been working since 2017 to provide outreach health care services to victims of major limb loss during conflict.

尽管我们的制图技术和空间数据能力有所进步, 但世界范围内的制图不平等却没有减弱。制图不平等体现在覆盖范围、质量和流通上, 在发展中国家的边远地区尤其严重。这些边远地区, 是世界上经济和资源最落后的社会, 需要高质量的制图来辅助配送重要的服务（如医疗服务）, 以面对贫困、冲突、全球气候变化等严峻挑战。自发地理信息（VGI）已经展现了消除制图不平等的潜力。VGI主要贡献于城市区域和应急反应（如地震、洪涝）, 但对承受着慢性人道危机的农村区域的关注度不够。一种替代解决方法是：更有效地发挥自发制图的作用, 为最需要的地区提供服务。机器学习就具有这种潜力。本文介绍的半人马VGI, 是结合了人类志愿者的空间认知能力和机器的速度和效率的一个混合系统。我们认为, 半人马VGI可以缓解VGI制图覆盖不公平中的一些政治和科技因素。文章以乌干达北部Acholi地区为例。该地区缺乏制图, 而作者从2017起就为该地区在冲突中的肢体伤残者提供医疗服务。

A pesar de los avances registrados en tecnologías cartográficas y de las capacidades de los datos espaciales, las desigualdades globales en mapeo no están declinando. Persisten las desigualdades en cobertura, calidad y aceptación del mapeo, con brechas significativas en las partes alejadas y rurales del Sur Global. Estas regiones, que representan algunas de las sociedades del mundo con mayores desventajas en términos de economía y recursos, requieren cartografía de alta calidad que contribuya a la provisión de servicios esenciales, salud, por ejemplo, en respuesta a retos severos tales como los de pobreza, conflicto y cambio climático global. La información geográfica voluntaria (VGI) ha mostrado potencial como solución a las desigualdades por mapeo. Sin embargo, en gran medida las contribuciones al respecto se han emprendido en áreas urbanas, o como respuesta a emergencias agudas (e.g., terremotos o inundaciones), dejando insuficientemente cartografiadas las regiones rurales, que sufren de crisis humanitarias crónicas. Se necesita una solución alternativa que aproveche el poder del mapeo voluntario de manera más efectiva para actuar sobre las regiones más necesitadas. El aprendizaje con máquinas es en cierta manera prometedor. En este artículo proponemos la VGI centauro, un sistema híbrido que combina las habilidades cognitivas espaciales de voluntarios humanos con la rapidez y eficiencia de la máquina. Sostenemos que la VGI centauro puede contribuir a mitigar algunos de los factores políticos y tecnológicos que generan desigualdades en la cobertura del mapeo VGI, y lo hacemos en el contexto de un estudio de caso en Acholi, en el norte de Uganda, una región inadecuadamente cartografiada en la que los autores han estado trabajando desde 2017 para suministrar servicios de salud de emergencia a quienes han sido víctimas de la pérdida de una extremidad principal durante el conflicto.

Key Words:

• humanitarian mapping
• machine learning
• mapping inequalities
• OpenStreetMap
• VGI

关键词:

• ：人道主义制图
• 机器学习
• 制图不平等
• OpenStreetMap
• 自发地理信息。

Palabras clave::

• aprendizaje con máquina
• desigualdades de mapeo
• mapeo humanitario
• OpenStreetMap, VGI

Acknowledgments

We thank Donna Sherman and Nick Scarle for their invaluable help in locating and digitizing colonial maps of Uganda, the Huckathon team for enabling the mapathons to take place, and the SEED Social Responsibility team at the University of Manchester for their support. We gratefully acknowledge the host of authors of the various open source software projects (referenced throughout this article) on which this project relies. Most of all, we thank the Acholi people for their kindness, generosity, and friendship.

Notes

1 Volunteered geographic information refers to the widespread voluntary engagement of private citizens in the creation of online geographic information through technologies such as crowdsourced mapping, Web 2.0 and social media, Global Positioning System, and smartphones (Haworth 2016).

2 Machine learning refers to computer systems that are capable of developing their own algorithms to solve complex problems (e.g., identifying features in imagery) by analyzing input data (e.g., satellite imagery) and desired result (e.g., predefined features) for a given problem. This contrasts with traditional computing approaches, where the algorithm is provided to the computer to derive results from input data (Domingues 2017).

3 A complete set of the maps is available in the Map Collection at the University of Manchester. A digital “slippy map” version covering the Acholi subregion is available at http://huckg.is/uganda50k.

Additional information

Funding

This work was supported by the AHRC/MRC GCRF Global Public Health Scheme under Grant AH/R005796/1 and a UKRI GCRF Global Impact Accelerator Grant.

Notes on contributors

Jonathan J. Huck

JONATHAN J. HUCK is a Lecturer in Geographical Information Science in the Department of Geography at the University of Manchester, Manchester M13 9PL, UK. E-mail: [email protected]. He is also Honorary Professor of Geographical Information Science at Gulu University, Uganda. He is interested in the application of maps and emergent technologies to geographical problems, particularly in the areas of health and conflict. In 2017 he founded Community Mapping Uganda (http://communitymapping.co.uk) to produce maps and provide map-based humanitarian services in northern Uganda.

Chris Perkins

CHRIS PERKINS is an Honorary Reader in Geography at the University of Manchester, Manchester M13 9PL, UK. E-mail: [email protected]. He has taught at Manchester since 1998, having previously run the university’s map libraries. His research interests lie at the interface between mapping technologies and social and cultural practices, with ongoing research into performative aspects of contemporary mapping behavior, an interest in sensory mapping, and an emerging interest in play.

Billy T. Haworth

BILLY T. HAWORTH is a Lecturer in Geographical Information Systems (GIS) and Disaster Management in the Humanitarian and Conflict Response Institute (HCRI) at the University of Manchester, Manchester M13 9PL, UK. E-mail: [email protected]. He teaches international disaster management and critical studies of GIS. His research interests include critical GIScience, participatory mapping, marginality and vulnerability, volunteered geographic information, citizen science, queer geographies, graffiti, and spatial knowledge production.

Emmanuel B. Moro

EMMANUEL B. MORO is an Associate Professor of Surgery and Former Dean of the Faculty of Medicine at Gulu University, Gulu, Uganda. E-mail: [email protected]. He is a Fellow of the College of Surgeons of East, Central and Southern Africa. He is interested in social transformation through educational and health research and innovations.

Mahesh Nirmalan

MAHESH NIRMALAN is Vice Dean for Social Responsibility and Public Engagement and Professor of Medical Education and Training in the Faculty of Biology, Medicine and Health at the University of Manchester, Manchester M13 9PL, UK. E-mail: [email protected] He is also Honorary Professor of Post Conflict Studies at Gulu University, Uganda. He is interested in healthcare research and delivery in post-conflict and low-resource settings.