IRC-Bench is a new dataset and evaluation framework for implicit entity recognition in reminiscence narratives, where entities must be inferred from non-local contextual cues across 1,994 transcripts linked to 12,337 WikiData entities.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
IRC-Bench: Recognizing Entities from Contextual Cues in First-Person Reminiscences
IRC-Bench is a new dataset and evaluation framework for implicit entity recognition in reminiscence narratives, where entities must be inferred from non-local contextual cues across 1,994 transcripts linked to 12,337 WikiData entities.