Paper 1 — Dual-Helix Governance for Agentic AI in WebGIS

“A Dual-Helix Governance Approach Towards Reliable Agentic Artificial Intelligence for WebGIS Development”
Levente Juhász, Boyuan Guan & Wencong Cui
Accepted in Transactions in GIS · ArXiv preprint

This paper is co-authored with Boyuan Guan and Wencong Cui, former colleagues from the FIU GIS Center, and represents a significant step forward in making agentic AI systems reliable enough for production geospatial engineering.

WebGIS development requires consistency, yet agentic AI often fails due to LLM context constraints, forgetting, stochasticity, instruction failure, and adaptation rigidity. We propose a dual-helix governance framework reframing these as structural problems rather than capacity deficits. Using a 3-track architecture (Knowledge, Behavior, Skills) and a persistent knowledge graph, it stabilizes execution by externalizing facts and enforcing protocols. Validation shows a governed agent successfully refactored a legacy WebGIS codebase (reducing cyclomatic complexity and improving maintainability), roughly halved trial-to-trial output variance relative to static prompting in a controlled experiment, and prevented common infodemic mapping errors in a 5-condition COVID-19 cartography ablation study. Operationalized via the open-source AgentLoom toolkit, this externalized governance provides the stability necessary for production-level geospatial engineering.

Paper 2 — DGGS Hash-Joins at Scale (ACM SIGSPATIAL ’26)

“Accelerating Point-in-Polygon Predicates via Algebraic Hash-Joins and Discrete Global Grids at Scale: An Interactive Benchmark”
Levente Juhász
Accepted at ACM SIGSPATIAL 2026, Riverside, CA · ArXiv preprint

This solo-author paper will be presented at ACM SIGSPATIAL ’26 later this year, one of the premier venues for spatial computing research.

Traditional vector-based point-in-polygon queries rely on computationally expensive geometric predicates that scale poorly for massive datasets, even when accelerated by spatial indices. Discrete Global Grid Systems (DGGS) offer a scalable alternative by discretizing geometries into hierarchical cells, transforming complex spatial relations into constant-time relational hash-joins. However, adopting a DGGS introduces an overhead to encode data, and current grid implementations exhibit a significant performance “tooling gap.” In this demonstration, we present an interactive dashboard that empirically evaluates these computational tradeoffs across four DGGS implementations (H3, S2, A5, and ISEA4H) using DuckDB. Through progressive scenarios, the platform visualizes the overhead of on-the-fly encoding and demonstrates how pre-indexing spatial datasets eliminates this overhead. Ultimately, the demo proves that when data is pre-indexed, all DGGS regardless of their mathematical complexity or tooling converge to sub-second join latencies, unlocking the throughput of modern vectorized execution engines.

A Maturing Research Agenda

These two acceptances are not coincidental. They represent the two most exciting frontiers in modern GIScience (well, at least to us), and both are core to what the GATOR Lab is building. Discrete Global Grid Systems are reshaping how we think about spatial data infrastructure at planetary scale, replacing expensive geometric predicates with algebraic efficiency, among other benefits. Agentic AI is redefining what it means to build geospatial software, raising fundamental questions about reliability, governance, and trust. We are just gettings tarted. Stay tuned.

Invited Talks in Europe — Milano and Szeged

The summer kicked off with a trip to Europe in June for two invited research talks.

At the Politecnico di Milano in Italy, I had the opportunity visit GEOlab (Geomatics and Earth Observation laboratory) led by Prof. Maria Antonia Brovelli to present our ongoing work on autonomous spatial intelligence. Founded in 1863, the Politecnico is one of the premier technical universities in Europe. I have been long following the work of Prof Brovelli’s group and meeting them in person was very rewarding. Thank you for the hospitality!

From Italy, the trip continued to Szeged, Hungary, where I was invited to speak at the University of Szeged (Szegedi Tudományegyetem). This was a full circle moment as my geospatial career started at the University of Szeged with degrees in Physical Geography and Geoinformatics! Both talks centered on the same theme — “Toward Autonomous Spatial Intelligence” — exploring how the convergence of large language models, discrete global grid systems, and agentic AI is reshaping the geospatial landscape.

These European visits were a great reminder of how broadly this research community spans, and how much appetite there is internationally for rigorous GeoAI research. Meeting new friend and collaborators, and reconnecting with old friends was definitely worth the travel.

The First GeoAI International Conference — Ghent, Belgium

One of the highlights of the summer was attending and presenting at the inaugural GeoAI 2026 Conference in Ghent, Belgium in early June. As the first dedicated international conference for Geospatial Artificial Intelligence, it brought together researchers and industry professionals working at the frontier of spatial data science, LLMs, and autonomous GIS from around the world.

I presented a joint study on LLM spatial data generation, which I co-authored with Paddy Gorry and Peter Mooney from Maynooth Univesity Ireland. This paper builds on Paddy’s PhD work and investigates how well large language models can generate geographic data — and where they fall short. Read the paper in the proceedings here. Being part of the launch of GeoAI as a recognized conference venue feels significant. This is a community coming into its own, and the enthusiam of the attendees was real. I look forward to future editions of this conference.

IJGI Webinar: GeoAI — Prospects and Challenges

On June 30, I had the pleasure of participating in a webinar organized by the ISPRS International Journal of Geo-Information (IJGI), titled “GeoAI – Prospects and Challenges.”. The was created to bring together experts to discuss the latest developments, applications, and future directions of Geospatial AI. It was a fitting venue — given my recent recognition as the 2026 IJGI Young Investigator — to engage directly with the journal’s community. The session was small, and only featured one other talk by Junghwan Kim, Director of the Smart Cities for Good Lab at Virginia Tech whose doing some great work in the GeoAI domain.

UCGIS Symposium — College Park, Maryland

Later in the summer, I attended the UCGIS Symposium in College Park, Maryland, the annual gathering of the University Consortium for Geographic Information Science. This is a community I value deeply for its commitment to advancing GIScience education and research in the USA.

In addition to attending, I had the opportunity to co-host a hands-on workshop at the Symposium. The session titled “From Prompts to Protocols: Govering Agentic AI for Reliable Geospatial Programming” focused on geospatial web development using governed agents. This workshops was co-hosted with my long-time collaborators from Florida International University, Dr. Boyuan Guan and Dr. Wencong Cui, and was based on our joint publication titled “A Dual-Helix Governance Approach Towards Reliable Agentic Artificial Intelligence for WebGIS Development”, which was just accepted for publication (literally today) in Transactions in GIS so stay tuned for the final publication! ]https://arxiv.org/abs/2603.04390

South Florida GIS Expo Workshop — West Palm Beach

Back in Florida, I led a pre-expo training workshop at the South Florida GIS Expo in West Palm Beach on August 19. The session, “AI’s Impact on the Future of GIS,” was a two-hour hands-on experience for GIS practitioners, with live demonstrations using a Google Colab notebook, applying semantic place classification to real data from the West Palm Beach area using LLM-powered tools built on top of the UF NaviGator Toolkit API.

It was a practical, applied session — the kind of event I find particularly rewarding because it connects academic research directly to the professional geospatial community. In turn, I get to learn about what GIS folks are interested in these days, which is always a useful input for training the next generation of GIS professionals. The South Florida GIS community is active and engaged, and the Expo is always a great venue for that kind of dialogue.

FSMS Event — Miami

Rounding out the summer closer to home, I also participated in an event organized by the Florida Surveying and Mapping Society (FSMS) in Miami, presenting at the Miami-Dade chapter meeting hosted at Miami Dade College. These local engagements are important — they keep the lab connected to the broader Florida geospatial profession and create pathways for students and practitioners to engage with cutting-edge research.

Looking Ahead

It has been a genuinely full summer. Across all of these events, a few themes kept surfacing: the growing recognition of GeoAI as a distinct and important subfield, the critical importance of spatial data quality, and the need for infrastructure — both technical and conceptual — that makes geospatial AI geometrically rigorous rather than geometrically naive.

The GATOR Lab is heading into Fall 2026 with new students, active manuscripts, and a lot of momentum. More to come soon.

Media and Resources

• Official News Release: Researcher’s pioneering work in AI and mapping earns international Young Investigator Award (UF/IFAS News)
• In-depth Interview: Interview with Dr. Levente Juhász (MDPI) — Discussion on the future of generative AI in geospatial contexts and advice for early-career researchers.

GIS Hour at Miami-Dade College

First, on Monday, April 13, I had the pleasure of presenting at the 2026 GIS Hour at Miami-Dade College, an event supported by ASPRS. I spoke about the increasing focus on GeoAI and Geospatial Analytics, and how modern Data Science is becoming more GIS-centered. It was a fantastic opportunity to connect with college students who are eager to explore geospatial careers.

Emerging Tech & Policy at Florida Atlantic University

Later in the month, on April 24, I participated in a high-impact workshop hosted by Florida Atlantic University’s Department of Civil, Environmental and Geomatics Engineering (CEGE) at the FAU Jupiter campus. The event, titled “Industry–Academic Collaboration on Emerging Tech & Policy and the Path to Professional Licensure,” brought together professionals and academics to discuss the future of the field.

I was honored to join a panel discussion focusing on the role of AI in geospatial contexts, and I also gave a presentation on discrete global grids. You can read more about the workshop and its outcomes here.

Bridging Theory and Application

The workshop featured inspiring keynotes from Michael Goodchild (UCSB), Shashi Shekar (UMN), May Yuan (UT Dallas) and Rahul Ramachandran (NASA), followed by lightning talks focused on computational infrastructure, scientific discovery, and the integration of social science within GeoAI. During the session I presented my position paper (available here) titled:

“Discrete Global Grid Systems as a Framework for Geometrically Rigorous and Spatially Explicit GeoAI: A Research Agenda”.

Much of the discussions over the two days emphasized the need for sharper integration between machine learning and spatial cognition, as well as the importance of community building to address both short- and long-term forecasting challenges.

Next Steps: Defining the Future of GeoAI

The energy from the meeting is already being channeled into a major perspective paper on the future of GeoAI. We are working to synthesize the collective insights from our discussions into a formal research agend. It was inspiring to see such a strong sense of shared identity emerging from this community.

A big thank you to Wenwen Li, Dylan Connor, Patricia Solis, Heather Baier, Daoqin Tong, Drew Trgovac, Peter Rogerson et al. the entire organizing committee for such a seamless and inspiring forum. I look forward to contributing to the perspective paper and shape the future of GeoAI together.

A History of Collaboration

This appointment builds on a long-standing relationship with IJGI. My previous contributions to the journal include:

• 2019 Outstanding Reviewer Award: Recognized for excellence and dedication to the peer-review process.
• April 2020 Cover Story: My research about Cartographic Vandalism was featured as the cover story for Volume 9, Issue 4.
• Special Issue Editor: I have served as a Guest Editor for several collections:
  • Advances in Social Network Analysis – Spatio-Temporal and Semantic Methods
  • OpenStreetMap as A Multi-Disciplinary Nexus: Perspectives, Practices and Procedures
  • Advances in AI-Driven Geospatial Analysis and Data Generation
  • Advances in AI-Driven Geospatial Analysis and Data Generation (2nd Edition)

Safeguarding Scientific Integrity

As an advocate for open science and reproducibility, I am committed to fostering open scientific exchange while maintaining high-quality research standards. My focus will be on ensuring that research is sound and validated, which is essential for reducing research waste and accelerating progress in geospatial analytics. I look forward to working with the editorial team to safeguard the integrity of the scholarly record in our field.

Why use this?

Selecting a publication venue is often based on “gut feeling” or a single metric like Impact Factor. This tool allows for a more refined, multi-dimensional assessment of the field:

• Meta Rankings: Compare journals using a composite index of eight different metrics (IF, CiteScore, SJR, SNIP, etc.) to see a more stable picture of “prestige”.
• Open Access Trends: Monitor the shift towards OA
• Collaboration Networks: Interactively explore how jurisdictions (countries) collaborate. Our data shows that while the field is growing, the Global South remains underrepresented… a gap we hope to highlight and help close to build an inclusive community.
• Cost vs. Impact: View the relationship between Article Processing Charges (APCs) and journal rankings to find the best “value” for your research.

Built for Open Science

In the spirit of open principles, the dashboard is built using Shinylive. This means it runs entirely in your web browser via WebAssembly, which in turn requires no backend server or 3rd party resources. It is private, and permanently accessible. The initial Shiny dashboard is a reimplementation and expansion of the 2024 IJGIS article and it was built by Jorge Corcino. Stay tuned as we are releasing the updated dataset and the source code of the dashboard very soon.

A word to the wise

We want to be clear about our view on using journal metrics to evaluate research impact. Quite frankly, we are not a fan of them. However, we need to be realistic and understand that it is a common practice in the academic community to use them. Think about going up for promotion, demonstrating the impact of your work, or justifying funding applications. Most of these decisions rely on some sort of metrics, and journal metrics are a common choice, so in a sense they are a necessary evil. While we may fundamentally disagree with putting to much emphasis on journal metrics, a lot of us are still at the mercy of them and therefore we are releasing this dashboard to build a better, more holistic understanding of the GIScience journal landscape.

On the bright side, better approaches to assess research impact are rapidly emerging, and we hope to see more of them in the future. A good resource to get started is the San Francisco Declaration on Research Assessment that recognizes and formalizes the need to improve the ways in which the outputs of scholarly research are evaluated.

With all its flaws, we still hope that our GIScience journal assessment dashboard will serve as a useful resource for everyone from students selecting their first submission venue to senior researchers optimizing publication strategies to maximize the impact of their work.

Explore the dashboard here or launch the Full Screen version.

Why GATOR?

The GATOR acronym is a play on words. It stands for Geospatial Analytics, Technology and Open Research. It represents the Geospatial (or Geomatics), Analytics, Technology and Open Research pillars of my research. As an alumnus, and now faculty member I also found it important to pay tribute to the Florida Gators🐊, and ground our lab’s mission within The Gator Nation.

What’s ahead?

I am currently in the process of setting up our core infrastructure. The next and ongoing major step is to recruit. I am currently seeking two fully funded PhD students to join the lab. This is an exciting opportunity not only for me, but for the first few lab members, who will be an important part of establishing the lab’s culture and direction. I am not just looking for the usual candidates, but for people who are passionate about open science and eager to solve complex spatial problems through a data-science lens. While an excellent academic background is important, I am more interested in your intellectual curiosity and a critical mindset. I can’t stress the importance of this enough. Finding a mutual fit will make or break this experience. Committing to a long PhD program is a major decision. I want to commit to your future success as much as the success of the GATOR Lab, therefore, I am very intentional and selective about who I recruit. I expect you to be the same with making sure that the GATOR Lab and my mentorship are a good fit for you. With that being said, I am interested in the person behind the CV, and I am open to discussing your interests and aspirations. If you are interested in joining the lab, please consult the join page for more information.

It is my hope that the lab will extend beyond PhD students. In the future, I plan to recruit postdocs, visiting scholars, and collaborators to push the boundaries of geospatial science. I am also interested in working with industry partners, and I am open to discussing potential collaborations, which will benefit the lab in many ways: it will inform our research, teaching, and help us stay at the cutting edge of geospatial analytics and technology.

Where are we located?

We are located in sunny South Florida, at the Fort Lauderdale Research & Education Center. While not being located in Gainesville may seem unusual, it is a great location for research and collaboration. South Florida (Miami - Ft. Lauderdale - West Palm Beach) is a vibrant metropolitan area. It is a great place to be, and it also provides a great environment for research and collaboration: it is a hub of innovation and creativity, with a diverse population and a rich cultural heritage. It also lies at the forefront of man-made and environmental challenges waiting to be solved by science.