measure-relevance

name	measure-relevance
description	Compute NDCG, MRR, and precision@k against the judgment list with a baseline. Reach for this before any tuning.

Skill: Measure relevance

'These results look better' is one session, not a metric (§3 #1).

Step 1 — Load the judgment list

Graded relevance labels for a representative query set (§3 #3).

Step 2 — Compute NDCG

DCG = Σ rel_i / log2(i+1) (rank i from 1); NDCG = DCG/IDCG via search_relevance_engineering_calc.py relevance (§3 #1).

Step 3 — Compute MRR + precision@k

MRR for first-relevant; precision@k for top-k quality (§3 #1).

Step 4 — Baseline before tuning

Record the baseline; every change reports a before/after delta (§3 #1 #3).

Output

An NDCG/MRR/precision@k read against the judgment list with a baseline. Traverse Tree 1 in the decision-trees file.