More_Like_This - Score Algorithm

Elastic Stack Elasticsearch
1

When using MLT Query it appears that it is ranking fields that match the names of the fields over fields with the actual terms.

Example:

I'm searching from a Prospect Type document that has a Name Field and a Company field. The name is a persons name "Davy Jones" and the Company Name is "Lockers Argh Us"

In my documents I have two other types. People and Companies.

People has a Name field which is the name of the People.
Companies has a Name field which is the name of the Company.

When I do MLT based on those fields. It will rank Jones Delicatessen over "Lockers Argh Us" even though it is an exact match, because Jones is in the Name field.

Any way around this?

2

Did some testing.

Boosting Company Field / Setting not_analyzed on Company. Neither worked.

It really wants to score Name in Name way over Company in Name.

No matter what boost i choose. It want's Davy's cycles or Jones Bar and Grill over Lockers or Us or anything to do with that.

I would love a way to say.. treat Company as if it was the Name field.