Let's say I am comparing two individuals, each with a first name, last name, postal code, address(line1), address(line2), and phone number. These all have varying reliability and importance for determining a match.
I can generate a J-W distance for each field with no issue. How can I combine these to get a sensible rating for the similarity of two individuals?
What I have thought of so far is just to weigh the different JW ratings to an amount proportionate to their relevance, and then take a weighted average. I feel like there must be some better techniques.