"confidence-score" scope and cardinality

[stephenbanghart0]

Question

Discussion from TWG meeting on 6/17/2025

The mapping model should include a “confidence-score” value that maps (from 0.0-1.0) how confident the author is that the given mapping is fully correct.

The precise mechanism for determining this value is left open for innovation. However, the Foundation/NIST may publish non-normative guidance.

Decision choices:

• 1 Scope:

  • 1A: “confidence-score” should be added as a top-level field for the entire mapping document.
  • 1B: “confidence-score” should be added as a field on each individual mapping.
  • 1C: “confidence-score” should be added as a field for BOTH the top-level document and each individual mapping.
    • Each value must accurately describe the confidence level of its scope – further specification (formulas, etc.) are left open for innovation.

• 2 Cardinality:

  • 2A: “confidence-score” should be a mandatory field.
  • 2B: “confidence-score” should be an optional field.

[xee5ch]

The mapping model should include a “confidence-score” value that maps (from 0.0-1.0) how confident the author is that the given mapping is fully correct.

How would any author perform scoring consistently, even if subjective, so one or more authors compare against one another?

The precise mechanism for determining this value is left open for innovation. However, the Foundation/NIST may publish non-normative guidance.

Given how subjective it is, what would non-normative guidance for subjective (and thus possibly inconsistent) confidence scores look like? When you have different decimal values for measurable values with objective comparison, that makes sense. How can you compare, sum, or perform other mathematical computations on subjective values that do not reasonable have the same "unit" of measurement?

In my opinion, this seems infeasible and easily gamed if the goal is objective measurement of mapping, so I give it at a 👎 vote.