Kvíz spojený s průvodcem · Střední úroveň

Entity Resolution for Financial Crime Quiz

Tests entity-linking purpose, false merges, match thresholds, graph limits, evaluation metrics, and investigator review.

Související vodicí cestyEntity Resolution For Financial Crime
Otázka 1 z 8

What does entity resolution attempt to determine?

Náhled každé otázky8 otázky

Odpovědi a vysvětlení se zobrazují, jak budete v kvízu.

  1. What does entity resolution attempt to determine?
    • Whether a transaction is illegal
    • Whether two names look identical only
    • Whether an account should be closed
    • Whether records refer to the same real-world person or organization
  2. Why are false merges harmful?
    • They improve recall without cost
    • They remove stale addresses
    • They can combine unrelated people and lead to unfair scrutiny
    • They always reveal criminal networks
  3. Which risk can arise from transitive graph errors?
    • All links become exact identifiers
    • One mistaken link can join many unrelated records into a cluster
    • Graph methods cannot represent transactions
    • A correct edge disappears
  4. What should an analyst inspect in a match?
    • Source identifiers, confidence, and why the records were linked
    • Only the final cluster color
    • Pouze název modelu
    • Only the number of accounts
  5. How can a system handle uncertain matches?
    • Auto-link every pair
    • Drop all low scores silently
    • Treat graph proximity as proof
    • Route them for review and preserve the possibility of no match
  6. Which metrics assess entity resolution?
    • Pairwise precision, recall, cluster purity, and downstream effects
    • Only model training time
    • Only number of records processed
    • Only number of graph nodes
  7. What can a shared address indicate?
    • Proof of money laundering
    • A guaranteed identity match
    • A possible relationship that needs additional evidence
    • Proof of common ownership
  8. How should an incorrect entity link be handled?
    • Treat the cluster as final evidence
    • Provide a correction process and preserve review history
    • Leave it in all downstream reports
    • Delete the original records