What is preference modeling?

Definition

Preference modeling tries to infer or encode what a person or group favors among alternatives. Inputs may include direct choices, rankings, stated principles, or other evidence, and the resulting model can be used to compare possible outcomes.

Inferred preferences are uncertain and sensitive to the data and assumptions used. A model built from synthetic personas cannot be treated as a validated picture of voters or public values without representative evidence and independent checks.

ELI5

Preference modeling is a structured guess about which options someone would choose. It turns examples or stated priorities into a pattern that can be tested against new choices.

For example, a system might compare a person's stated data-sharing preferences with a contract and highlight where they disagree. The person still decides whether the inferred preference is accurate.

Frequently asked questions

What evidence can preference modeling use?

It can use explicit choices, rankings, pairwise comparisons, stated rules, and carefully justified contextual evidence.

Why are inferred preferences uncertain?

People's choices depend on context, available options, changing priorities, and evidence that a model may not observe.

Videos explaining preference modeling

  1. Brendan Rappazzo in a plain blue T-shirt beside the headline Stress Test Your Morals on a pure black background.