Pipeline Distinction Loss Index (PDLI)

GMI-EGD-023Evidence-Governed DataInfrastructure-DependentPattern O-19: Cross-Model Interaction

Definition

The share of claims where a distinction present in the source evidence and required by a downstream resolver, grader, or rule was no longer present after upstream extraction and normalization.

The Measurement Gap

There is no metric that tells you whether an upstream normalization or extraction step destroyed a distinction a downstream resolver or grader required, which means every component tests sound in isolation and the pipeline degrades the claim.

Formula

PDLI = D_lost / D_required

Proposed formula, pending validation pilots.

At a Glance

Unit
Ratio (0 to 1)
Trend Direction
Lower
Category
Lane Routing and Rule Certification

Decision Supported

Whether normalization and extraction settings are tuned for their own accuracy at the cost of the pipeline's.

← Forensic Population Narrowing Rate (FPNR)Rule Specification Validation Rate (RSVR) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment