Testing the Quality of GeoAI-Generated Data for VGI Mapping

Dear Colleagues,

Geospatial Artificial Intelligence (GeoAI) trains models on large spatial datasets to discover patterns and make predictions for various purposes. For example, it can use these models to generate, update, and analyse geographic information, fundamentally reshaping how maps are produced, maintained, and consumed. Compared to traditional (still largely manual) methods, GeoAI-generated data can quickly create detailed maps of built environments (e.g., roads, buildings), which are essential for urban planning, disaster management, environmental monitoring, and many other location-based services and downstream applications.

While several prestigious GIScience journals have recently published Special Issues to highlight GeoAI’s remarkable potential as a tool for automated mapping, none have explicitly focused on the rising problem of “AI slop” in this domain. To address this problem, this Special Issue presents the latest research findings that assess the quality of contemporary GeoAI model outputs in terms of map feature accuracy and reliability, with the goal of maintaining VGI maps. AI data quality/reliability is a concern to all VGI/crowd-source mappers. For example, the OpenStreetMap community follows a “code of conduct” to control the uploading of automated edits into the live OSM database; this is a well-known policy to both VGI practitioners and GIScience researchers alike.

In this context, GeoAI data quality refers to both the accuracy and reliability of model-predicted map features. Where accuracy relates to its real-world position, orientation, and shape - to answer the question: Is GeoAI ready to take over? While reliability is more subjective, referring to a model’s consistency/trustworthiness, e.g., does it pass the eye-test - to answer the question: Are we ready to let it?

Previous studies published in IJGI and elsewhere have demonstrated that an accuracy/consistency gap exists between VGI/crowd-source map features and various authoritative “ground truth” spatial datasets, and that modern GeoAI solutions could potentially bridge this gap.

To explore this claim, this IJGI Special Issue is especially interested in empirical GeoAI studies from GIScience researchers around the world to attest the quality, in terms of accuracy and reliability, of their model outputs (map feature predictions) when compared directly to either/both authoritative (e.g., national mapping agency) “ground truth” data and/or current VGI maps (e.g., OSM) within their respective regions. The aim is to inform today’s VGI mapping community of the current state-of-the-art regarding GeoAI accuracy and inherent reliability, or not, for keeping crowd-source maps up-to-date.

• Applications of GeoAI to online mapping;
• Use cases integrating GeoAI data into VGI maps;
• Empirical accuracy and reliability evaluations of GeoAI map feature predictions;
• Benchmarking frameworks for GeoAI;
• Investigations of QA metrics for GeoAI-generated data;
• Investigations of QA metrics for crowd-source data;
• Future directions/challenges of GeoAI for mapping;

Dr. James D. Carswell
Dr. Lasith Niroshan
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. ISPRS International Journal of Geo-Information is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1900 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• GeoAI
• segmentation
• deep learning
• automated mapping
• OpenStreetMap (OSM)
• volunteered geographic information (VGI)