# Entity resolution with Elasticsearch, part 4: The ultimate challenge

DevFeed: [Entity resolution with Elasticsearch, part 4: The ultimate challenge](<https://devfeed.tech/articles/entity-resolution-with-elasticsearch-part-4-the-ultimate-challenge-78823.md>)

Original publisher: [Read original article](<https://www.elastic.co/search-labs/blog/entity-resolution-elasticsearch-llm-challenges>)

Author: Jessica Moszkowicz

Published: 2026-03-13T00:00:00Z

Content type: article

Language: en

Sources: [Elasticsearch Labs](<https://devfeed.tech/sources/elasticsearch-labs.md>)

Topics: [Entity resolution](<https://devfeed.tech/topics/entity-resolution.md>), [elasticsearch](<https://devfeed.tech/topics/elasticsearch.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [audit](<https://devfeed.tech/tags/audit.md>), [basics](<https://devfeed.tech/tags/basics.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [elasticsearch](<https://devfeed.tech/tags/elasticsearch.md>), [entity-resolution](<https://devfeed.tech/tags/entity-resolution.md>), [hybrid-search](<https://devfeed.tech/tags/hybrid-search.md>)

## AI overview

The article evaluates a search first entity resolution pipeline that uses Elasticsearch to retrieve candidate matches and an LLM to judge them through function calling. A diverse dataset with 51 challenge categories produced about 91% precision, 86% recall, and 89% F1, with weaker performance in some multilingual cases attributed largely to limitations in the dense vector model. The author presents the architecture as an educational prototype and notes that production use would need further validation and safeguards.

## Source excerpt

Solving and evaluating entity resolution challenges in a highly diverse "ultimate challenge" dataset designed to prevent shortcuts.