Semantic record matching

Phase 1: recall

SearchCluster builds indices on your record’s data that allow for broad searches, with phonetic variations, simplifications.

Your search is transformed internally to hit these indices, and find candidate matches, records that have enough in common to be considered for further analysis.

Phase 2: match

The candidate records are loaded from data store, decrypted in the process, and compared entity by entity, attribute by attribute, with the search input. That’s the work of our Matcher component. The detailed match results are passed to the caller.