Publication
Do LLMs fail in bridging generation?
Natalia Skachkova; Simon Ostermann; Josef van Genabith; Bernd Kiefer
In: Ngoc Duyen Tanja Tu; Annelen Brunner; Christian Lang (Hrsg.). Journal for Language Technology and Computational Linguistics (JLCL), Vol. 38, No. 2, Pages 77-95, ACL Anthology, 7/2025.
Abstract
In this work we investigate whether large language models (LLMs) ‘understand’ bridgingrelations and can use this knowledge effectively. We present the results obtained fromtwo tasks: generation of texts containing bridging and filling in missing bridging spans.We show that in most of the cases LLMs fail to generate bridging in a reliable way.