---
title: "What is entity resolution?"
description: "Entity resolution defined: linking records that refer to the same real-world entity, and how it differs from identity and coreference resolution."
canonical: https://past.dev/glossary/entity-resolution
last-updated: 2026-09-02
---
# What is entity resolution?

Source: https://past.dev/glossary/entity-resolution

Entity resolution is identifying that different records refer to the same real-world entity and linking them under one identity. The records may come from different systems: a CRM row, an email sender, a meeting transcript, a support ticket. Resolution decides whether they describe one entity or several, then links them so facts about that entity accumulate in one place instead of fragmenting across partial copies.

## How entity resolution works

Entity resolution takes records from different systems and decides which of them describe the same entity. The entity can be a person, a company, a project, or any subject that appears in more than one place.

1. **Candidate generation.** Group records that could plausibly match, using keys such as normalized names, email domains, or shared attributes. This step limits how many pairs need comparison.
2. **Matching.** Compare candidate pairs. Deterministic rules link records that share a hard identifier, such as the same email address. Probabilistic scoring handles softer signals, such as name similarity plus a shared employer.
3. **Clustering.** Merge matched records under one stable identity, and record the links so a wrong merge can be reversed later.

This page defines the term. The [entity resolution guide](/guides/entity-resolution) covers matching rules and failure modes in depth, and the [testing guide](/guides/testing-entity-resolution) shows how to verify a resolver before trusting its links.

## Entity resolution vs identity resolution vs coreference resolution

The three terms are frequently confused. They operate at different scopes.

| Term | Operates on | Example |
| --- | --- | --- |
| **Entity resolution** | Records across systems | A CRM row for "Nordwind GmbH" and an invoice from nordwind.io become one company |
| **[Identity resolution](/glossary/identity-resolution)** | One person's identifiers across channels | jane@nordwind.io, the chat handle @janek, and user id 4411 become one person |
| **[Coreference resolution](/glossary/coreference-resolution)** | Mentions inside one text | In "Jana presented and she took questions", the pronoun links to Jana |

A memory pipeline typically runs them in order: coreference resolution cleans up each document, entity resolution links subjects across documents and systems, and identity resolution maintains the person-level graph. The [identity resolution guide](/guides/identity-resolution) covers the person case separately.

## Entity resolution in agent memory

An agent's memory ingests the same subjects through many channels. On March 3 a support ticket arrives from "J. Alvarez". On March 12 an email arrives from jalvarez@nordwind.io. On April 2 a meeting transcript mentions "Jorge from Nordwind". Without resolution, memory holds three unconnected subjects, and a question about Jorge returns only the transcript.

With resolution, the three records link to one person, and recall over any of the identifiers returns the full history. Facts accumulate on the entity instead of scattering across partial profiles. past.dev runs entity resolution during ingestion, linking references to the same person or subject across email, transcripts, tickets and chat; see the [quickstart](/docs/memory-api/quickstart).

## Related concepts

- **[Identity resolution](/glossary/identity-resolution)**: the person-focused case, linking one individual's identifiers across channels.
- **[Coreference resolution](/glossary/coreference-resolution)**: linking mentions inside a single text before records are compared.
- **[Memory consolidation](/glossary/memory-consolidation)**: the background pass where entity resolution runs after ingestion.
- **[Memory graph](/glossary/memory-graph)**: the structure resolved entities and their dated facts are stored in.

## Frequently asked questions

### What is the difference between entity resolution and identity resolution?

Entity resolution links records that refer to the same real-world entity of any kind, including companies and projects. Identity resolution is the person-focused case: connecting one individual's identifiers across channels such as email, chat, and application accounts.

### Why do AI agents need entity resolution?

Agent memory ingests the same people and companies through many channels. Without resolution, facts about one customer split across several partial profiles, and recall returns an incomplete history.

### Is entity resolution the same as deduplication?

Deduplication removes exact or near-exact copies of records. Entity resolution goes further: it links records that look different but refer to the same entity, such as a nickname in a transcript and a formal name in a CRM.

## Related

- [Entity resolution guide](https://past.dev/guides/entity-resolution)
- [Testing entity resolution](https://past.dev/guides/testing-entity-resolution)
- [What is identity resolution?](https://past.dev/glossary/identity-resolution)
- [What is coreference resolution?](https://past.dev/glossary/coreference-resolution)
- [Benchmark methodology](https://past.dev/benchmarks/methodology)