To attempt to clean results and merge multiple data sources together, we make use of a name normalization algorithm. This algorithm takes input such as “mr john smith jr” and turns it into “Smith, Mr John, Jr”. Other examples might be taking “Smith Al” and determining if it should be “Smith, Al” or “Smith, A L”. Name normalization also helps us determine that “Smith, A L” is the same as “Smith, AL”, so we don’t show multiple names on a record unnecessarily.
Today, we’ve released a significant improvement to this algorithm. With it, we’re hoping to see cleaner results for everyone.
We’re also considering releasing an API for this algorithm. If that sounds like a good idea to you, let us know!