Are there better algorithms for computing list similarity that this one I wrote, or corrections?

Your example name looks to be Spanish. I know nothing about that.

In English names

• first names are often reduced to nicknames, so if you are searching for Richard Brown, you would need to search your document for Dick Brown.

• multiple names ( e.g. "Jose Alfredo Plaza Navarro ) are often reduced to two. The reduction is not random. The last name is usually kept. Some first names are more often kept than others ( e.g. Richard Alphonse Brown and Alphonse Richard Brown will BOTH become Richard Brown.

• Last name is most often kept, preceded by a title and an abbreviation. Richard Brown will often be seen as Mr R Brown. You will also need to include Mr. R. Brown and similar - suggest you strip out any '.' before doing any comparisons.

I expect Spanish names will show different but equivalent modifications.

"I want the windows with the most common elements with respect to the target name to score higher "

I think this is a mistake. You should not penalize a missing middle name so much.

What are you intending to do about false positives?