Bug Fixes
Duplicate Person Records Now Collapse
People whose profiles were split across multiple records now resolve to a single record instead of several. This usually happened when someone had two LinkedIn URLs, or conflicting job histories on the same identity. Previously, a conflicting signal on one part of a profile could prevent an otherwise well-evidenced match from being made, leaving the same person present two or more times with differentlegion_ids.
If you store legion_id, you will need to reconcile. When duplicate records collapse, one legion_id survives and the others are retired. A lookup on a retired legion_id will no longer return a record. Re-resolve any stored ids against the current data.
This build collapsed 224,538 duplicate identities into 222,280 surviving records. Most collapses merged a single pair, and the largest merged seven records into one.
The vast majority of ids are unaffected: 186,081,935 legion_ids carried forward unchanged.
Titles Recognized as Name Prefixes
Titles such as Pastor, Rabbi, Brother and Imam are now recognized as name prefixes and populated inprefix. Where the record also carries a given name, that name is used for first_name instead of the title. Name suffixes that aren’t recognized (anything outside jr, sr, and ii–x) are dropped rather than passed through as-is.
Names legitimately beginning with St are unaffected.
Certification Parsing
Certification names that use unusual bullet characters or non-breaking spaces are now parsed correctly rather than retaining the leading punctuation incertifications[].name.