Leveraging Grammatical Framework and WordNet for Natural Language Generation from Wikidata
Paper in proceeding, 2025
We present a prototype for the generation of Wikipedia articles from Wikidata in multiple languages. For the generation we leverage the multilingual resources in the GF Resource Grammars Library (RGL), the GF WordNet and the existing labels in Wikidata. The result is a system where we get an out-of-the-box baseline generation for currently 26 languages which can be quickly adjusted by hand to produce publishing quality documents. We evaluate six of the languages.
Author
Krasimir Angelov
University of Gothenburg
Undergraduate Education (GRU)
Andrea Carrion del Fresno
Chalmers, Computer Science and Engineering (Chalmers), Functional Programming
University of Gothenburg
Ekaterina Voloshina
University of Gothenburg
Chalmers, Computer Science and Engineering (Chalmers), Functional Programming
Aarne Ranta
Chalmers, Computer Science and Engineering (Chalmers), Computing Science
University of Gothenburg
Lecture Notes in Networks and Systems
23673370 (ISSN) 23673389 (eISSN)
Vol. 1198 LNNS 173-1849783031764585 (ISBN)
21st International Symposium on Distributed Computing and Artificial Intelligence, DCAI 2024
Salamanca, Spain,
Salamanca, Spain,
Subject Categories (SSIF 2025)
Natural Language Processing
Computer Sciences
Comparative Language Studies and Linguistics
DOI
10.1007/978-3-031-76459-2_16