Brief Taxonomy & Scoring Guide

Per Story Scores

Confidence Score: Overall trust score (0-100), computed in backend.

Relevance Score: How relevant the story is for the brief (0-100).

Relevance Confidence: Model confidence in its relevance judgment (0-100).

Evidence Strength: Quality and strength of supporting evidence (0-100).

Narrative Certainty: How stable and clear the narrative looks (0-100).

Polarization Score: How polarized the framing is (0-100). Lower is better for confidence.

Formula

confidence_score =
0.30 * relevance_confidence
+ 0.30 * evidence_strength
+ 0.20 * narrative_certainty
+ 0.10 * (100 - polarization_score)
+ 0.10 * relevance_score

Color Dots

Green: High score quality.

Amber: Medium score quality.

Red: Low score quality.

Tab Configuration

Each card shows the current ingestion source set and allowed broad tags.