Entity extraction identifies and classifies named items in text: individuals, companies, providers, dates, monetary amounts, medications, locations, and document references. It is often the first structured layer built from raw documents.
Entity resolution follows, determining that references such as Dr. Smith, R. Smith MD, and the treating orthopedist point to the same person.
Alternative Names:
Named Entity Extraction, Entity Recognition
Why it Matters?
Resolution is the harder problem and the one that determines usefulness. Medical records refer to the same provider a dozen ways, and productions spell party names inconsistently across custodians. A system that extracts entities but does not resolve them produces a list that still requires manual reconciliation, which is most of the work it was supposed to save.
Frequently Confused with
Related terms
Frequently asked questions
What is entity resolution?
Why is resolution harder than extraction?
