Search Results

Documents authored by Liu, Zilong

Document
DOI: 10.4230/LIPIcs.COSIT.2026.7
LMM Modality Transfer: A Pre-Requisite for Autonomous GIS Agents

Authors: Ivan Majic, Zexian Huang, Franziska Hübl, Krzysztof Janowicz, Meilin Shi, Mina Karimi, Zilong Liu, and Alexandra Fortacz-Lazan

Published in: LIPIcs, Volume 393, 17th International Conference on Spatial Information Theory (COSIT 2026)

Abstract
AI models are becoming increasingly adept at understanding and processing spatial information, thereby facilitating agentic problem-solving in spatial tasks and workflows. However, most of the research on their spatial capabilities (e.g., spatial reasoning) has focused on the textual modality as input and output. This contrasts with the human approach to GIS workflows, where text and visual modalities are often used together, interchangeably, and in a complementary manner. Thus, to truly achieve an automated GIS analysis pipeline or carry out human-designed GIS workflows, AI models - Large Multimodal Models (LMMs) in particular - need to be able to seamlessly transition between image- and text-based modalities that are traditionally used in such workflows. We present a modality transfer task that (1) asks an LMM to first describe an input image of colored squares in a regular grid, and (2) asks a new LMM instance to re-generate an image of the original spatial scene using the textual description output by the former model. This task quantifies the ability of LMMs to transfer spatial information between image and text modalities. Ultimately, by examining the modality transfer capability of LMMs through the lens of spatial information theory, this work highlights a critical bottleneck: achieving strong and robust geospatial understanding in LMMs requires rigorous, multi-modal alignment. Our results indicate that recent LMMs (here from OpenAI) still struggle with modality transfer, when tasked with re-generating an image of a simple spatial grid of color squares.
Cite as

Ivan Majic, Zexian Huang, Franziska Hübl, Krzysztof Janowicz, Meilin Shi, Mina Karimi, Zilong Liu, and Alexandra Fortacz-Lazan. LMM Modality Transfer: A Pre-Requisite for Autonomous GIS Agents. In 17th International Conference on Spatial Information Theory (COSIT 2026). Leibniz International Proceedings in Informatics (LIPIcs), Volume 393, pp. 7:1-7:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)

Copy BibTex To Clipboard

@InProceedings{majic_et_al:LIPIcs.COSIT.2026.7,
  author =	{Majic, Ivan and Huang, Zexian and H\"{u}bl, Franziska and Janowicz, Krzysztof and Shi, Meilin and Karimi, Mina and Liu, Zilong and Fortacz-Lazan, Alexandra},
  title =	{{LMM Modality Transfer: A Pre-Requisite for Autonomous GIS Agents}},
  booktitle =	{17th International Conference on Spatial Information Theory (COSIT 2026)},
  pages =	{7:1--7:20},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-438-3},
  ISSN =	{1868-8969},
  year =	{2026},
  volume =	{393},
  editor =	{Timpf, Sabine and Filomena, Gabriele and Kapaj, Armand and Zhu, Rui and Giudice, Nicholas A. and Manley, Ed},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.COSIT.2026.7},
  URN =		{urn:nbn:de:0030-drops-275514},
  doi =		{10.4230/LIPIcs.COSIT.2026.7},
  annote =	{Keywords: Large Multimodal Model (LMM), Spatial Reasoning, GIS Agent, Modality Transfer, GeoAI}
}
Document
Short Paper
DOI: 10.4230/LIPIcs.COSIT.2026.21
Conceptual Vagueness in Human and LLM Representations: Evidence from Migration Distance Categorization (Short Paper)

Authors: Songlin Wang, Krzysztof Janowicz, Ailin Benitez Cortes, Marion Borderon, Yingjing Huang, Mina Karimi, Zilong Liu, Annika Süß, and Patrick Sakdapolrak

Published in: LIPIcs, Volume 393, 17th International Conference on Spatial Information Theory (COSIT 2026)

Abstract
Conceptual vagueness is a fundamental characteristic of many geographic categories. The varied definitions of "long-distance" and "short-distance" in migration studies provides a typical example. Although interpretations of these categories are inherently context-dependent and heterogeneous, they remain widely used in domain research. This study investigates the extent to which different individuals reach a consensus on conceptualizing these categories. With a migration related corpus where "long" and "short" are generally not explicitly mentioned, we elicit human and LLM judgments in classifying each article into these two categories. Both human-human and human-LLM agreements are quantified. Our results show that human agreements are only moderate, confirming substantial conceptual variability among informed readers. LLM outputs exhibit a comparable level of variability, failing to converge on a stable category boundary. This disagreement arises from multiple sources, from detection of implicit signals to their interpretations. These findings demonstrate the feasibility of using LLMs as probes to characterize conceptual vagueness in geography at scale. Returning to migration distance, we highlight that they should not be treated as universally shared categories in knowledge organization and representation. In contrast, resolving such vagueness requires richer contextual information or additional semantic signals.
Cite as

Songlin Wang, Krzysztof Janowicz, Ailin Benitez Cortes, Marion Borderon, Yingjing Huang, Mina Karimi, Zilong Liu, Annika Süß, and Patrick Sakdapolrak. Conceptual Vagueness in Human and LLM Representations: Evidence from Migration Distance Categorization (Short Paper). In 17th International Conference on Spatial Information Theory (COSIT 2026). Leibniz International Proceedings in Informatics (LIPIcs), Volume 393, pp. 21:1-21:9, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)

Copy BibTex To Clipboard

@InProceedings{wang_et_al:LIPIcs.COSIT.2026.21,
  author =	{Wang, Songlin and Janowicz, Krzysztof and Benitez Cortes, Ailin and Borderon, Marion and Huang, Yingjing and Karimi, Mina and Liu, Zilong and S\"{u}{\ss}, Annika and Sakdapolrak, Patrick},
  title =	{{Conceptual Vagueness in Human and LLM Representations: Evidence from Migration Distance Categorization}},
  booktitle =	{17th International Conference on Spatial Information Theory (COSIT 2026)},
  pages =	{21:1--21:9},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-438-3},
  ISSN =	{1868-8969},
  year =	{2026},
  volume =	{393},
  editor =	{Timpf, Sabine and Filomena, Gabriele and Kapaj, Armand and Zhu, Rui and Giudice, Nicholas A. and Manley, Ed},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.COSIT.2026.21},
  URN =		{urn:nbn:de:0030-drops-275652},
  doi =		{10.4230/LIPIcs.COSIT.2026.21},
  annote =	{Keywords: qualitative categories, conceptual vagueness, human migration, large language models, agreement}
}
Document
DOI: 10.4230/LIPIcs.GIScience.2025.16
What, When, and Where Do You Mean? Detecting Spatio-Temporal Concept Drift in Scientific Texts

Authors: Meilin Shi, Krzysztof Janowicz, Zilong Liu, Mina Karimi, Ivan Majic, and Alexandra Fortacz

Published in: LIPIcs, Volume 346, 13th International Conference on Geographic Information Science (GIScience 2025)

Abstract
Inundated by the rapidly expanding AI research nowadays, the research community requires more effective research data management than ever. A key challenge lies in the evolving nature of concepts embedded in the growing body of research publications. As concepts evolve over time (e.g., keywords like global warming become more commonly referred to as climate change), past research may become harder to find and interpret in a modern context. This phenomenon, known as concept drift, affects how research topics and keywords are understood, categorized, and retrieved. Beyond temporal drift, such variations also occur across geographic space, reflecting differences in local policies, research priorities, and so forth. In this work, we introduce the notion of spatio-temporal concept drift to capture how concepts in scientific texts evolve across both space and time. Using a scientometric dataset in geographic information science, we detect how research keywords drifted across countries and years using word embeddings. By detecting spatio-temporal concept drift, we can better align archival research and bridge regional differences, ensuring scientific knowledge remains findable and interoperable within evolving research landscapes.
Cite as

Meilin Shi, Krzysztof Janowicz, Zilong Liu, Mina Karimi, Ivan Majic, and Alexandra Fortacz. What, When, and Where Do You Mean? Detecting Spatio-Temporal Concept Drift in Scientific Texts. In 13th International Conference on Geographic Information Science (GIScience 2025). Leibniz International Proceedings in Informatics (LIPIcs), Volume 346, pp. 16:1-16:18, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2025)

Copy BibTex To Clipboard

@InProceedings{shi_et_al:LIPIcs.GIScience.2025.16,
  author =	{Shi, Meilin and Janowicz, Krzysztof and Liu, Zilong and Karimi, Mina and Majic, Ivan and Fortacz, Alexandra},
  title =	{{What, When, and Where Do You Mean? Detecting Spatio-Temporal Concept Drift in Scientific Texts}},
  booktitle =	{13th International Conference on Geographic Information Science (GIScience 2025)},
  pages =	{16:1--16:18},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-378-2},
  ISSN =	{1868-8969},
  year =	{2025},
  volume =	{346},
  editor =	{Sila-Nowicka, Katarzyna and Moore, Antoni and O'Sullivan, David and Adams, Benjamin and Gahegan, Mark},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.GIScience.2025.16},
  URN =		{urn:nbn:de:0030-drops-238450},
  doi =		{10.4230/LIPIcs.GIScience.2025.16},
  annote =	{Keywords: Concept Drift, Ontology, Large Language Models, Research Data Management}
}
Document
Short Paper
DOI: 10.4230/LIPIcs.COSIT.2024.23
Towards Formalizing Concept Drift and Its Variants: A Case Study Using Past COSIT Proceedings (Short Paper)

Authors: Meilin Shi, Krzysztof Janowicz, Zilong Liu, and Kitty Currier

Published in: LIPIcs, Volume 315, 16th International Conference on Spatial Information Theory (COSIT 2024)

Abstract
In the classic Philosophical Investigations, Ludwig Wittgenstein suggests that the meaning of words is rooted in their use in ordinary language, challenging the idea of fixed rules determining the meaning of words. Likewise, we believe that the meaning of keywords and concepts in academic papers is shaped by their usage within the articles and evolves as research progresses. For example, the terms natural hazards and natural disasters were once used interchangeably, but this is rarely the case today. When searching for archived documents, such as those related to disaster relief, choosing the appropriate keyword is crucial and requires a deeper understanding of the historical context. To improve interoperability and promote reusability from a Research Data Management (RDM) perspective, we examine the dynamic nature of concepts, providing formal definitions of concept drift and its variants. By employing a case study of past COSIT (Conference on Spatial Information Theory) proceedings to support these definitions, we argue that a quantitative formalization can help systematically detect subsequent changes and enhance the overall interpretation of concepts.
Cite as

Meilin Shi, Krzysztof Janowicz, Zilong Liu, and Kitty Currier. Towards Formalizing Concept Drift and Its Variants: A Case Study Using Past COSIT Proceedings (Short Paper). In 16th International Conference on Spatial Information Theory (COSIT 2024). Leibniz International Proceedings in Informatics (LIPIcs), Volume 315, pp. 23:1-23:8, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2024)

Copy BibTex To Clipboard

@InProceedings{shi_et_al:LIPIcs.COSIT.2024.23,
  author =	{Shi, Meilin and Janowicz, Krzysztof and Liu, Zilong and Currier, Kitty},
  title =	{{Towards Formalizing Concept Drift and Its Variants: A Case Study Using Past COSIT Proceedings}},
  booktitle =	{16th International Conference on Spatial Information Theory (COSIT 2024)},
  pages =	{23:1--23:8},
  series =	{Leibniz International Proceedings in Informatics (LIPIcs)},
  ISBN =	{978-3-95977-330-0},
  ISSN =	{1868-8969},
  year =	{2024},
  volume =	{315},
  editor =	{Adams, Benjamin and Griffin, Amy L. and Scheider, Simon and McKenzie, Grant},
  publisher =	{Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
  address =	{Dagstuhl, Germany},
  URL =		{https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.COSIT.2024.23},
  URN =		{urn:nbn:de:0030-drops-208386},
  doi =		{10.4230/LIPIcs.COSIT.2024.23},
  annote =	{Keywords: Concept Drift, Semantic Aging, Research Data Management}
}

Any Issues?
X

Feedback on the Current Page

CAPTCHA

Thanks for your feedback!

Feedback submitted to Dagstuhl Publishing

Could not send message

Please try again later or send an E-mail
© 2023-2026 Schloss Dagstuhl – LZI GmbH Schloss Dagstuhl – LZI GmbH About DROPS Imprint Privacy Contact