Can Tisane help me analyze, validate, or compare names?
Yes.
Parsing Full Names
We parse the names into constituents:
- given_name
- middle_name
- surname
- title
- suffix
- social_role
To parse a name, invoke /parse, adding "entity":"person","words":true
into the settings parameter. The constituents of the name are marked with the role
attribute, as shown on the screenshot below. If found, the full name is encapsulated into an entity structure (under entities_summary
).
Real Name Validation
We have no means of knowing whether a person provided their real name. However, we can recognize many names of important people, historic figures, spiritual beings, imaginary characters, or names that don’t appear to be names, such as the ones described in this Quora post. The entity structure in the /parse method’s response contains a subtype attribute with the following values (Wikidata ID is provided alongside):
- important_person
- fictional_character
- spiritual_being
Comparison
Tisane’s /compare/entities method allows comparing two names in the same or different languages and returning a list of components that differ. For example:
- William Smith vs Will Smith returns
{"result":"different","differences":["variation"]}
, - Musa Bin Osman vs Haji Musa Bin Osman returns
{"result":"different","differences":["social_role"]}
- William Smith vs Вилл Смит (Will Smith in Russian) returns
{"result":"different","differences":["variation"]}
- Kevin Tan vs TAN Kevin returns
{"result":"same"}
Vetting Nicknames & Aliases
Nicknames, just like other types of user-generated content, are prone to abuse. Most online communities do not welcome usernames like Hitler, UserJohn_is_liar, etc.
To vet a nickname or an alias, invoke /parse with "format":"alias"
. The subscope
setting determines whether to attempt segmenting the string for languages normally using white space (e.g. when it’s delimited by underscores, camel case, or just written without spaces).
To sign up to Tisane API (free plan with up to 50,000 requests per month and no time limit), follow this link.