# Named-Entity Recognition

> NER finds spans of text that name entities and assigns them types such as people, organisations, products, or identifiers.

- HTML version: https://robbiepalmer.me/ideas/ner
- Source: https://en.wikipedia.org/wiki/Named-entity_recognition

Named-entity recognition, or NER, finds mentions of entities in text and assigns each mention a
type. A general model may recognise people, organisations, and places. A domain model might instead
need shipment numbers, carriers, ingredients, medical conditions, or account identifiers.

NER is an [NLP](/ideas/nlp) task, but its labels come from the problem being solved. The same text
can need different boundaries and types in different systems. "New York Times" may be one
organisation mention in a news index, while another task may care about the place name inside it.
[Ontology Engineering](/ideas/ontology-engineering) makes those categories and distinctions
explicit.

Recognition does not establish identity. Finding two mentions of "Starbucks" does not prove that
they refer to the same company or branch. Entity resolution handles that later question. In
[Intelligent Document Processing](/ideas/intelligent-document-processing), NER can turn recognised
text into candidate fields before validation and review.

## Questions it prompts

* Which entity types affect a real product decision?
* How should overlapping, nested, or ambiguous mentions be labelled?
* Does evaluation score exact text boundaries as well as the entity type?
* Which rules, dictionaries, or checks should complement the model?

## Related ideas

- [Intelligent Document Processing](https://robbiepalmer.me/ideas/intelligent-document-processing.md): A pipeline that turns varied documents into validated, structured data by combining recognition, language understanding, and workflow rules.
- [NLP](https://robbiepalmer.me/ideas/nlp.md): Natural language processing uses computational methods to analyse, transform, and generate human language.
- [Ontology Engineering](https://robbiepalmer.me/ideas/ontology-engineering.md): The practice of defining the concepts, relationships, constraints, and shared vocabulary needed to represent a domain.

## Where it appears

- Project: [Intelligent Document Processing](https://robbiepalmer.me/projects/intelligent-document-processing.md)
- Project: [Automated PII Detection](https://robbiepalmer.me/projects/pii-detection.md)
- Blog post: [How To Automatically Detect PII for Real-Time Cyber Defense](https://robbiepalmer.me/blog/2023-05-23-automatically-detect-pii-real-time-cyber-defense.md)

---

Markdown index of this site: https://robbiepalmer.me/llms.txt