Extracting spatial relations from unstructured text

Authors: Chen-Chieh Feng*, Geography, National University of Singapore
Topics: Geographic Information Science and Systems, Spatial Analysis & Modeling
Keywords: relations, semantics, GIS, digital scholarship
Session Type: Virtual Paper
Day: 4/7/2021
Start / End Time: 11:10 AM / 12:25 PM
Room: Virtual 46
Presentation File: No File Uploaded


Unstructured text has recently emerged as a source of rich spatial information, especially in the field of digital humanity given the long history of textual documentation on, for example, human activities, local environments, and their interactions. Extracting spatial information from unstructured text and formalize it is an important enabler to mine such information and gain insight into the past, which in turns provide the foundation to understand long-term changes of human-environment interactions. Previous study along this this line has relied mainly on rule-based approaches to extract spatial relations from text. While useful, they tend to be limited by scalability, given the need of handcrafting these rules for specific text sources, and the lack of consideration of semantics. This study will explore how existing machine learning approaches may improve spatial relation extraction, using a text documenting the biography of prominent Singaporean Chinese as a case study. The presentation will compare the performance of using existing machine learning tools for the stated purpose and how they may be improved for automatic extraction of spatial relations, which will support digital humanity applications.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login