[Improvement]: Location Disambiguation \ Human Name Matching - Rule

[Jared Van Valkengoed]

This might not be ideal but feel free to address concerns. We will need to check for words after. (Most likely need to drop proper noun at end & expand last name list - or will return things like "Toronto Maple" as person).

function nameRule(str) {
  // Create an NLP document from the input text
  let doc = nlp(str);
  // Check for patterns like  "Orlando Bloom" or "Paris Hilton"
  let match = doc.match("(#Place+|#Place|#ProperNoun|#Noun) (#LastName|#Person|#ProperNoun)");
  if (match.found) {
    return match.out("text");
  }
 

  return false; // Return null if no name is found
}


console.log(nameRule("I went to Orlando today to visit Orlando Bloom")) 
console.log(nameRule("I went to Paris today and seen Paris Hilton")) 

This should tag things like "Orlando Bloom" only as a "person" same way compromise handles "Paris Hilton" and does not return a place.

Now only if we can solve handling names like "North West" ie: Kayne / Kim's kid as a person and not directions without a neural network. 😿

[spencer kelly]

You're free to do this in your own project!

[Jared Van Valkengoed]

@spencermountain - do you not think this might not be useful for determining if a entity is a name or location properly...? As said tho - the last proper noun should be dropped and the last name list should be expanded.

[spencer kelly]

That would tag all consecutive Propernouns as people. Please try it on some text, and see how it works, before proposing a change to this project.

[Jared Van Valkengoed]

@spencermountain - I am going to sound stupid (nor trying to sound arrogant) but can I have some examples of some strings so I can see what you mean?

And if not a rule already maybe just #Place #LastName and expand last name list?