Beneficial owner, four documents, four name spellings

Name variants in corporate registry documents are a common and often underestimated obstacle in financial intelligence work. A beneficial owner listed as "Oleksandr Koval" in one filing appears as "Alexander Koval" in a second, "A. Koval" in a third and "Aleksander Kowal" in a fourth. Each variant is consistent with the same individual: transliteration differences, an anglicised first name, an abbreviated initial, a Polish spelling of a Ukrainian surname. But one of them could also be a different person, and conflating the two is an analytical error that compounds downstream.

The manual approach is to treat each variant as a separate entity until positive evidence links them, then build the evidence set: same date of birth, same passport number, same address, a photograph match. That is correct procedure but it is slow, and it does not scale when an analyst is working across dozens of beneficial ownership records with name variants across multiple languages and transliteration systems. The cognitive load of tracking which variants have been resolved and which remain open typically exceeds what a single session can hold.

The problem is further complicated by the fact that some variants are discriminating and some are cosmetic. "Oleksandr" versus "Alexander" is cosmetic: Ukraine's official romanisation against the anglicised form of the same name, with no investigative weight. "Kowal" versus "Koval" is potentially discriminating. The two share the same root, both meaning "blacksmith", but "Kowal" is the Polish spelling and "Koval" the romanisation of the Ukrainian. The difference may be nothing more than a translator's convention, or it may indicate a Polish national who is a different person entirely. Knowing which variants to investigate and which to collapse requires linguistic knowledge the analyst may not have for every language pair in the document set.