Entity resolution matches variant references to a single underlying entity, handling name variations, abbreviations, titles, typographical errors, and role descriptions. It typically combines string similarity with contextual signals such as dates and associated facts.
Output is a canonical entity with all its references linked.
Alternative Names:
Record Linkage, Entity Matching
Why it Matters?
Resolution is where extraction becomes usable. Medical records refer to the same physician as Dr. Smith, R. Smith MD, the orthopedist, and Smith, Robert across different forms, and a system that lists these as four providers has produced a chronology someone must still reconcile by hand. In productions, party and custodian name variants create the same problem at larger scale.
Frequently Confused with
Related terms
Frequently asked questions
Why is entity resolution harder than extraction?
What happens without it?
