Comp4321 The Index May 14 2026

The anatomy of a top result.

query · hkust admission · doc № 001
phrase ×3.0 title ×2.0 body match high pagerank

Forward index · top stems on this page

admiss 2 cse 2 depart 2 hkust 2 back 1 main 1 pg 1 ug 1

document length 12 body tokens  ·  max tf 2 (admiss)  ·  vocabulary 8 distinct stems  ·  title 3 stems

Score breakdown

Re-derived from the same tf·idf cosine values, boost factors, and PageRank multipliers the live ranker just used. weighted = body + title × 2.0 · final = weighted × phrase × pagerank.

  1. cosine(query, body) tf·idf body · 2 resolved stems0.6741
  2. cosine(query, title) × 2.0 title boost ×2.0 applied0.7764
  3. weighted subtotal cosBody + 2.0 × cosTitle1.4504
  4. × phrase multiplier ≥2 stems within 5-token window×3.0
  5. × pagerank multiplier 1 + 0.5 × normalized PR (0.007)×1.0037
  6. final score weighted × phrase × pagerank4.3675

Per-query-stem · how each term contributes

  1. hkust wid 0007 · df 4 body tf 2 title tf 1 idf 6.229
  2. admiss wid 0004 · df 3 body tf 2 title tf 0 idf 6.644

Outlinks (3)