The 5th International Workshop on Geographic Information Extraction (GeoExT) will be held at the 49th European Conference on Information Retrieval (ECIR’27) in Southampton, UK, March 21-25, 2027.

Website: https://geo-ext.github.io/GeoExT2027

Important dates

  • Submission deadline: January 17, 2027
  • Notification of acceptance: February 19, 2027
  • Camera ready: March 15, 2027
  • Workshop day: March 21-25, 2027

Scope

The workshop welcomes contributions in all topics related to geographic information extraction from texts, focused (but not limited) to:

  • Advances in toponym recognition, resolution, and document-level geocoding
  • Scalable and resource-efficient geographic information extraction
  • Advanced geocoding methods supporting spatial footprints and extents
  • Geographic information extraction for underrepresented languages, regions, and text types
  • Geographic biases, robustness, uncertainty, explainability, and reliability of GeoExT systems
  • Methods for interpreting historical and ancient toponyms in interdisciplinary contexts
  • Techniques for handling complex location descriptions in text
  • Geospatial knowledge graph construction from unstructured texts
  • Fusion of text-based geographic information with other spatial data sources
  • Development and annotation of datasets designed for the requirements of different applications
  • Developing high-quality gazetteers for underrepresented regions
  • Development and annotation of datasets designed for the requirements of different applications
  • Evaluation methodologies, benchmarks, reproducibility, error analysis, and uncertainty quantification for GeoExT
  • Novel applications showcasing the use of geographic information embedded in texts
  • LLMs, vision-language models and retrieval-augmented approaches for geographic information extraction, including their efficiency, reliability and geographic biases
  • Integration of extracted geographic information into information retrieval systems, including geographic search, spatial indexing, location-aware ranking, and geo-aware retrieval-augmented generation.

Submissions

Intended to support discussions and dialogues, we would like to encourage the following types of submissions:

  • Full paper (oral presentation): up to 8 pages
  • Short paper (oral presentation): up to 4 pages
  • Demonstration paper (hands-on demo): up to 4 pages

The references are not counted in the page limit for all the submission types. All manuscripts are expected to address the topics related to geographic information extraction from texts and should be original and unpublished contributions. Pre-prints submitted to ArXiv are eligible. Manuscripts should follow the CEUR-ART single-column style and be submitted as PDF files to EasyChair: https://easychair.org/my/conference?conf=geoext2025.

IMPORTANT: Please include between brackets the type of submission (full; short and oral; demo) in the paper title. Each manuscript will be peer-reviewed by at least two program committee members, following the single-blind paradigm. Accepted papers will be included in an open-access proceedings volume of CEUR Workshop Proceedings, indexed by both Scopus and DBLP.

Organizing committee

  • Xuke Hu, German Aerospace Center, Germany
  • Ludovic Moncla, INSA Lyon, France
  • Inhye Kong, University of Zurich, Switzerland
  • Christopher Jones, Cardiff University, UK