Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
October 3, 2025Open Access

Findings of the Fourth Shared Task on Multilingual Coreference Resolution: Can LLMs Dethrone Traditional Approaches?

View Full Paper
Ask AI
Bookmark
Share

Authors

MNMichal NovákMKMiloslav KonopíkANAnna Nedoluzhko

Discussion

Loading...

Member takes

Overview

Overview shows LLMs can challenge traditional systems in multilingual coreference resolution, indicating a shift in approaches.

Key Points

  • LLMs demonstrated potential in multilingual coreference resolution, indicating they could rival traditional methods in future tasks.
  • Nine systems participated, including four LLM-based approaches — two were fine-tuned and two used few-shot adaptation techniques.
  • The introduction of new datasets expanded the task's coverage, utilizing version 1.3 of CorefUD across 17 languages.
  • The simplified plaintext format aimed to optimize the task for LLMs versus the original CoNLL-U representation.

Cite This Study

Novák et al. (2025) studied this question.

synapsesocial.com/papers/68e0450fa99c246f578b4083https://doi.org/10.48550/arxiv.2509.17796
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Improving LLMs' Learning for Coreference Resolution2025
  2. 2The Generalization Ability of Coreference Resolution Systems2026
  3. 3LLMs4OL 2025 Overview: The 2nd Large Language Models for Ontology Learning Challenge2025 · 2 citations
  4. 4Do LLMs exhibit the same commonsense capabilities across languages?2025
  5. 5SCUNLP-1 at the NTCIR-18 FinArg-2 Task: Collaborative Large Language Models for Temporal Classification2025