Entity resolution finds records that describe the same real-world person, company, product, or place, even when they are stored differently. A customer may appear as Jon Smith in a billing system and as Jonathan Smith with a different email in a support tool. Entity resolution compares attributes such as names, addresses, identifiers, and dates, scores how likely two records match, and links or merges them. It is also called record linkage, and it underpins master data management.
Matching rules carry judgment. Two similar records can be one customer or two people who share an address, and two orders on the same day can be a duplicate or a repeat purchase. Teams therefore keep the original records and the match decisions, so a merge can be reviewed or reversed. For AI work, how records are linked changes what a model or agent treats as one customer, so the rules need to fit the task.