The first stage of coreference is mention detection, which is the task of finding the spans of text that constitute each mention. Mention detection often starts with all nouns and named entities and then uses anaphoricity classifiers or referentiality classifiers to filter out non-mentions

Mention Detection

Linguistic expressions like *her* or *Victoria Chen* that refer to some entity. Mentions are typically noun phrases, pronouns, or proper nouns.

Carleton College

The task of disambiguating whether or not two or more mentions corefer to the same discourse entity

Coreference Resolution

An on-going but a helpful book resource about NLP
https://web.stanford.edu/~jurafsky/slp3/

Speech and Language Processing (3rd ed. draft) 

Supervised neural machine learning systems are used to model coreference, mainly used coreference architectures are 

 - The Mention-Pair Architecture
 - The Mention-Rank Architecture
  - Entity-based Models


Architectures for Coreference Algorithms

Two or more referring expressions are said to corefer if they refer to the same discourse entity

Corefer

Mention

Entity linking is the task of mapping a discourse entity to some real-world individual

Entity Linking

These properties are often used to aid coreference resolution:

- Number Agreement
- Person Agreement
- Gender/Noun Class Agreement
- Binding Theory -> mentions within the same sentence typically require a reflexive pronoun such as *himself* to corefer
- Recency -> mentions are more likely to corefer with more recent utterances that older ones
- Grammatical role -> Entities mentioned in subject position are more salient than those in object position, which are in turn more salient than those mentioned in oblique positions
- Verb semantics -> semantic meaning of verbs can help bias the interpretation of subsequent pronouns

Learn Before

Related

Learn After