Phase 3: Transformers for NLP · ~60 minutes · Python

Coreference Resolution

"She called him. He did not answer. The doctor was at lunch." Three references to two people and nobody is named. Coreference resolution figures out who is who.

What you will learn

Implement coreference resolution from scratch
Understand the math and intuition behind the algorithm
Use production libraries for the same task
Ship a reusable artifact

Introduction

Type: Learn
Languages: Python
Prerequisites: Named Entity Recognition, POS Tagging and Syntactic Parsing
Time: ~60 minutes

The Problem

Extract every mention of Apple Inc. from a 300-word article. Easy when the article says "Apple." Hard when it says "the company," "they," "Cupertino's technology giant," or "Jobs's firm." Without resolving these mentions to the same entity, your NER pipeline misses 60-80% of the mentions.

Coreference resolution links every expression that refers to the same real-world entity into one cluster. It is the glue between surface-level NLP (NER, parsing) and downstream semantics (IE, QA, summarization, KG).

Why it matters in 2026:

Summarization: "The CEO announced..." vs "Tim Cook announced..." — the summary should name the CEO.
Question answering: "Who did she call?" requires resolving "she."
Information extraction: a knowledge graph with "PER1 founded Apple" and "Jobs founded Apple" as separate entries is wrong.
Multi-document IE: merging mentions across articles about the same event is cross-document coreference.

Unlock the full lesson

You've read the first 2 sections. The rest of this lesson covers The Concept, Build, Check Yourself, Key Terms & Next — plus a hands-on lab, quiz, and project artifact.

Create a free account — the opening phases of 29 of 35 courses are free, no credit card.