Coreference Resolution Explainer
Coreference resolution identifies expressions that refer to the same entity or concept. In a passage, a person’s name, a job title, and pronouns such as she or they may all belong to one coreference chain.
A system first finds candidate mentions and then scores whether pairs or clusters should be linked. Contextual language representations help compare distant mentions, while constraints can account for number, syntax, speaker, and document structure.
Coreference information improves information extraction, summarization, question answering, translation, and analysis of events across long documents. It lets downstream tools combine facts that use different wording for the same subject.
Pronouns can be genuinely ambiguous, and some references depend on common sense or conversation context. Datasets also differ in what counts as a mention. Errors can merge different people or split one entity, so evaluation should inspect both link and cluster behavior.
Coreference resolution determines which words or phrases in a passage refer to the same person, object, event, or idea.
Key terms include coreference, mention, antecedent, pronoun, coreference chain, cluster, anaphora, discourse and ambiguity.
Mentions are detected and grouped into chains representing the same entity or idea.
Context, syntax, and discourse help rank possible antecedents.
Grammar alone may not reveal which person or object a pronoun means.
Systems should preserve uncertainty when more than one interpretation remains plausible.
Explore more "Explainers"
Discover additional explainers across politics, science, business, technology, and other fields. Each explainer breaks down a complex idea into clear, everyday language—helping you better understand how major concepts, systems, and debates shape the world around us.
