Navigation auf uzh.ch
Geographic information is embedded in a wide range of text documents, from news articles and historical texts to social media posts, often expressed through place names (toponyms). Extracting and accurately grounding these toponyms is essential for various research applications, such as analysing the geographic contexts of historical narratives, understanding spatial patterns of discourses in social media, and exploring the geographic dimensions of literary works. Geoparsing is the task of automatically identifying and disambiguating these toponyms within texts and associating them with their correct geographical locations.
In this workshop, participants will be introduced to Geoparser, a Python library designed to efficiently geoparse large collections of text. The first part of the workshop will cover setting up Geoparser, exploring its features, and using it to extract and disambiguate toponyms in text documents. Through hands-on exercises, attendees will learn how to apply Geoparser to their own data.
The second part of the workshop will focus on customising Geoparser for domain-specific text corpora. Participants will learn how to prepare training data, fine-tune the model, and evaluate its performance, to improve the accuracy of geoparsing when working with texts from a specific domain. By the end of the workshop, attendees will have a solid foundation for using Geoparser and adapting it to their projects.
When / Wann: November 29, 2024, 09:00 - 12:00
Where / Wo: SOC-E-006 «Schrödinger», Digital Society Initiative, Rämistrasse 69, 8001 Zurich
Language / Sprache: English
More information / Weitere Infos: Link
Organisation: DSI GIS Hub