The repository accompanies our NAACL 2025 publication "ConMeC: A Dataset for Metonymy Resolution with Common Nouns"
Dataset Access:
The dataset can be downloaded directly from Hugging Face: https://huggingface.co/datasets/Rey97/ConMeC
What is ConMeC?
ConMeC is the first large-scale, systematically annotated dataset focused on common noun metonymy. It is designed to support tasks in metonymy resolution, figurative language understanding, and LLM evaluation.
Each instance includes:
• A full sentence
• The target noun
• A binary label indicating whether the noun is used metonymically (1) or literally (0)
• Metonymy category tags (container, producer, product, location, causer, possessed)
• Preceeding and suceeding sentences for additional context
• Document URL and title
Example
Target Word: stadium
Sentence: "he said as the stadium cheered, ""It's the bottom of the 7th, the last chance to order beer."
Label: Metonymic
Category: Location