Navigation auf zora.uzh.ch

Search ZORA

ZORA (Zurich Open Repository and Archive)

Maintenance: On Wednesday, February 12th 2025, maintenance work will take place on the MariaDB servers from 9:45pm to approx. 10:45pm. During this period, ZORA and JDB will be unavailable. Thank you for your understanding.

Fine-tuning transformers for toponym resolution: A contextual embedding approach to candidate ranking

Gomes, Diego; Purves, Ross S; Volpi, Michele (2024). Fine-tuning transformers for toponym resolution: A contextual embedding approach to candidate ranking. In: GeoExT 2024: Second International Workshop on Geographic Information Extraction from Texts at ECIR 2024, Glasgow (Scotland), 24 March 2024. CEUR-WS, 43-51.

Abstract

We introduce a new approach to toponym resolution, leveraging transformer-based Siamese networks to disambiguate geographical references in unstructured text. Our methodology consists of two steps: the generation of location candidates using the GeoNames gazetteer, and the ranking of these candidates based on their semantic similarity to the toponym in its document context. The core of the proposed method lies in the adaption of SentenceTransformer models, originally designed for sentence similarity tasks, to toponym resolution by fine-tuning them on geographically annotated English news article datasets (Local Global Lexicon, GeoWebNews, and TR-News). The models are used to generate contextual embeddings of both toponyms and textual representations of location candidates, which are then used to rank candidates using cosine similarity. The results suggest that the fine-tuned models outperform existing solutions in several key metrics.

Additional indexing

Item Type:Conference or Workshop Item (Paper), refereed, original work
Communities & Collections:07 Faculty of Science > Institute of Geography
Dewey Decimal Classification:910 Geography & travel
Scopus Subject Areas:Physical Sciences > General Computer Science
Language:English
Event End Date:24 March 2024
Deposited On:09 Jan 2025 14:24
Last Modified:10 Jan 2025 21:01
Publisher:CEUR-WS
Series Name:CEUR Workshop Proceedings
Number:3683
ISSN:1613-0073
OA Status:Green
Free access at:Publisher DOI. An embargo period may apply.
Official URL:https://ceur-ws.org/Vol-3683/paper6.pdf
Download PDF  'Fine-tuning transformers for toponym resolution: A contextual embedding approach to candidate ranking'.
Preview
  • Content: Published Version
  • Language: English
  • Licence: Creative Commons: Attribution 4.0 International (CC BY 4.0)

Metadata Export

Statistics

Downloads

2 downloads since deposited on 09 Jan 2025
2 downloads since 12 months
Detailed statistics

Authors, Affiliations, Collaborations

Similar Publications