How we score matches JSON

The OpenSanctions API supports matching of entities using a simple query-by-example mechanism. For transparency, you can find the weighting of features used in that API here.

The API uses a simple entity comparison model based on logistic regression. Both the training data and the code are fully public, inviting public scrutiny and proposals for improvement.

Matching features

Feature	Coefficient	Description
`name_match`	1.234	Check for exact name matches between the two entities.
`name_token_overlap`	0.075	Evaluate the proportion of identical words in each name.
`name_numbers`	-0.221	Find if names contain numbers, score if the numbers are different.
`name_levenshtein`	0.434	Consider the edit distance (as a fraction of name length) between the two most similar names linked to both entities.
`phone_match`	0.040	Matching phone numbers between the two entities.
`email_match`	0.034	Matching email addresses between the two entities.
`identifier_match`	0.214	Matching identifiers (e.g. passports, national ID cards, registration or tax numbers) between the two entities.
`dob_matches`	1.190	The birth date or incorporation date of the two entities is the same.
`dob_year_matches`	0.224	The birth date or incorporation year of the two entities is the same.
`first_name_match`	0.068	Matching first/given name between the two entities.
`family_name_match`	0.105	Matching family name between the two entities.
`birth_place`	-0.106	Same place of birth.
`gender_mismatch`	-0.261	Both entities have a different gender associated with them.
`country_mismatch`	-0.323	Both entities are linked to different countries.
`org_identifier_match`	0.547	Matching identifiers (e.g. registration or tax numbers) between two organizations or companies.
`address_match`	0.902	Text similarity between addresses.
`address_numbers`	0.218	Find if names contain numbers, score if the numbers are different.