STxFM Workshop 2026

About

As urban environments worldwide become increasingly interconnected and data-rich, it is crucial to explore the implications of integrating advanced artificial intelligence with complex physical-world data. This workshop aims to delve into the profound influence of Large Language Models (LLMs), and more generally, foundations models, the structure and dynamics of the spatio-temporal ecosystem. Specifically, we aim to foster research on the interplay between foundation models and critical urban elements, including points of interest (POIs), human mobility flows and trajectories, road network structures, and urban regions. By bringing together experts from natural language processing (NLP), data mining, and spatio-temporal database management, the workshop provides a platform to discuss ideas, review past work, and explore new directions. The focus is on the synergy between foundation models and spatio-temporal data from two primary aspects:

Foundation models-empowered Spatio-Temporal Engineering. Exploring how LLMs can empower tasks such as trajectory data cleansing, relation inference between spatial entities, and natural language interfaces for geographic query processing.
Spatio-Temporal Data-enhanced LLMs. Examining how structured spatial knowledge and mobility data can enhance LLMs through Retrieval-Augmented Generation (RAG) and multi-modal data management to address issues like hallucination and spatial reasoning.

By including participants from both academia and industry, the workshop aims to connect practical urban applications with cutting-edge research, encouraging collaborations that will shape the future of intelligent spatio-temporal systems and their impact on societal well-being.

Topics of Interest

Spatio-temporal data is critical in modern applications such as urban planning, autonomous driving, and climate monitoring. When the volume of trajectory and sensor data grows, traditional methods face challenges in complex pattern recognition and natural language interaction. The emergence of Large Language Models (LLMs) offers a transformative opportunity to redefine how we process, analyze, and query spatio-temporal information. This workshop explores the synergy between LLMs, or in a more general form, foundation models, and spatio-temporal data engineering, focusing on how LLMs can enhance spatio-temporal engineering and reasoning, and how structured spatio-temporal knowledge bases can ground LLMs in physical reality to reduce hallucinations. This workshop aims to unite researchers from GIS, database, and NLP communities.

The workshop welcomes theory and methodology papers falling into the scope of following themes, including but not limited to:

LLM-empowered Spatio-Temporal Data
- Trajectory Data Cleansing and Map-matching with LLMs
- Natural Language Interfaces for Spatial Databases (Text-to-SQL for GIS)
- Zero-shot and Few-shot Learning for Urban Flow Prediction
- LLM-based Spatial Reasoning and Navigation Instructions
- Interpretable Spatio-Temporal Forecasting via LLMs

Spatio-Temporal Data Enhanced LLMs
- Retrieval-Augmented Generation (RAG) with Geographic Knowledge Bases
- Tokenization Strategies for Continuous Spatio-Temporal Coordinates
- Pre-training LLMs with Multi-modal Spatio-Temporal Data
- Benchmarking LLMs on Spatial Intelligence and Geographic Logic
- Privacy-preserving Spatio-Temporal Data Synthesis using LLMs
- Urban foundation models

Submission Details

Accepted papers will be published in the MDM 2026 Workshop Proceedings and published by IEEE . Submitted papers must be original work that has not appeared in, and is not under consideration for, another conference or journal. After being assessed for suitability by the workshop Chairs, all submitted papers will be single-blind peer-reviewed by the Program Committee members. Every submitted paper will be reviewed by at least three members of the Program Committee.

Workshop papers may be up to 6 pages in the IEEE conference proceedings format.

All submissions must follow the IEEE Computer Society Proceedings Manuscript Formatting Guidelines.

The official templates are available at https://www.ieee.org/conferences/publishing/templates.

Submissions will be received via CMT https://cmt3.research.microsoft.com/STxFM2026.
Important : registering on the submission site with a title and a meaningful abstract by the earliest deadline is required to enable the actual paper submission.

Venue	📌 TBD
14:00 - 14:10	👏 Opening Remarks
14:10 - 14:50	🎯 Keynote from Academia
14:50 - 15:30	📄 Research Oral Presentation
15:30 - 15:50	☕ Tea Break
15:50 - 16:30	🎯 Keynote from Industry
16:30 - 16:40	🏆 Best Paper Award & Closing Remarks
16:40 - 17:30	🪧 Research Poster Session