liveBook · Manning

Coreference resolution

Coreference resolution is the task of identifying expressions in text that refer to the same entity and linking those expressions to it.

How 4 Manning books explain it

Natural Language Processing in Action, Second Edition

Coreference resolution is the task of determining which entity or expression a referring expression, such as a pronoun, denotes in text.

How it is used
Connects repeated entity mentions to consolidate information extracted into knowledge graphs.
How deep it goes
Defines the task, demonstrates ambiguous references, and shows resolver output without detailing algorithm internals.
Examples
Uses Maria and her pronouns, Timnit Gebru and she, and an ambiguous city-councilmen pronoun.
What you can do after
Apply it via Coreferee to process text and inspect the resulting coreference chains.

Read how Natural Language Processing in Action, Second Edition explains coreference resolution →

Graph Algorithms for Data Science

Coreference resolution is the task of identifying expressions in text that refer to the same entity and linking them to that entity.

How it is used
Supports information extraction, text summarization, and question-answering workflows by resolving references before downstream processing.
How deep it goes
Explains definitions, ambiguity, approach types, and an implementation, but stops short of training models or detailed optimization.
Examples
Uses Google and YouTube ambiguity, plural company references, and an Apple Inc. passage with deliberately exposed resolution errors.
What you can do after
Apply it via SpaCy and coreferee to replace references with resolved entities and inspect downstream extraction risks.

Read how Graph Algorithms for Data Science explains coreference resolution →

Rearchitecting LLMs

Coreference resolution is the task of determining which previously mentioned entity a referring expression, such as a pronoun, denotes.

How it is used
Uses WinoGrande coreference examples to calibrate block-importance analysis for depth pruning.
How deep it goes
Details dataset formatting, forward hooks, activation capture, and cosine-based block scoring.
Examples
Uses a sentence about Ian, Dennis, and menudo to illustrate choosing the entity referred to by a pronoun.
What you can do after
Evaluate model-block importance using activations captured on formatted WinoGrande examples.

Read how Rearchitecting LLMs explains coreference resolution →

Knowledge Graphs and LLMs in Action

Coreference resolution is the technique of determining which entity or expression a referring phrase, such as a pronoun or shortened name, denotes.

How it is used
Supports accurate text interpretation and knowledge-graph extraction by linking referring expressions to entities.
How deep it goes
Defines pronoun reference through an ambiguity example and notes shortened-name failures; stops short of specifying a resolution algorithm.
Examples
Uses the John–Bob pronoun ambiguity and archival abbreviations such as “S.” and “U.Cal.” as concrete cases.
What you can do after
Apply prompted GPT to resolve references and extract related entities and relations, while recognizing failures on shortened names.

Read how Knowledge Graphs and LLMs in Action explains coreference resolution →

Related concepts

Questions about coreference resolution

What is coreference resolution?

Coreference resolution is the task of identifying expressions in text that refer to the same entity and linking those expressions to it.

Which Manning books explain coreference resolution?

Natural Language Processing in Action, Second Edition, Graph Algorithms for Data Science, Rearchitecting LLMs and Knowledge Graphs and LLMs in Action.

What kinds of coreference resolution are there?

Kinds of coreference resolution include neural coreference resolution and rule-based coreference resolution.

What is coreference resolution part of?

Coreference resolution is part of information extraction. Information extraction is the computational process of identifying specified facts, entities, or relationships in input data and producing structured information.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.