Jev AI use cases
Data matching
Compare records across sources and separate likely matches from cases that need review.
Data matching
Compare records across sources and separate likely matches from cases that need review.
About data matching
Data matching decides whether records from different sources describe the same real-world entity: a product listed by two sellers, a company spelled two ways, a person in two systems. It is the comparison step inside entity resolution and record linkage.
Jev compares a candidate pair under the identity policy you write, and returns a same, different or review decision with field-level signals. Candidate generation, normalization, clustering and the final merge stay in your pipeline, where they can be audited.
Which page fits your task
- Entity MatchingYou have candidate record pairs, such as products or organizations, and need a match decision that respects variants, identifiers and ambiguous names.
Common mistakes
- Matching on similar names alone. Variants and identifiers usually decide identity.
- Treating a missing identifier as a conflict, or a conflict as a missing value.
- Assuming matches are transitive. Check the whole cluster before merging.
Data matching FAQ
What is the difference between entity matching and entity resolution?
Entity matching compares two records. Entity resolution is the whole process: normalize data, generate candidate pairs, compare them, resolve conflicts and maintain links. Jev handles the comparison step.
Will Jev merge or deduplicate my database?
No. It returns decisions for the pairs you send. Your system decides whether to link or merge, ideally after reviewing uncertain pairs.
Further reading · reviewed 2026-09-23
- Knowledge graph entity alignment cookbookTypeSafe documentation · docs.typesafe.ai/cookbooks/entity_alignment
- Categorical decisions with ChoiceTypeSafe documentation · docs.typesafe.ai/primitives/choice
One decision model. Your own rules.
Use a Noul for a yes/no probability, a Choice for a fixed set of outcomes, or a Score for an ordered rubric. Ask focused questions about the same input and combine the results in code.