Multilingual coreference resolution task expands datasets, tests LLMs

The Fifth Shared Task on Multilingual Coreference Resolution, held at the CODI-CRAC 2026 workshop, focused on systems capable of identifying mentions and clustering coreferential chains, with a particular emphasis on long-range entities. The task expanded to include five new datasets and two additional languages, utilizing version 1.4 of CorefUD, which now covers 27 datasets in 19 languages. Ten systems participated, including four LLM-based approaches, and while traditional systems still led, LLMs showed considerable promise for future editions. AI

IMPACT LLMs show promise in coreference resolution, potentially challenging traditional methods in future tasks.

RANK_REASON The cluster is based on a research paper detailing a shared task and its findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• CODI-CRAC 2026
• CorefUD
• LLM

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →