✿ Gangnam
Beauty GuideReview evidence
Interactive prototype · illustrative data

Know what sits behind the score

Trust should be inspectable.

A working evidence layer that makes review sources, translation, likely duplicates and verification visible before a patient compares clinics.

Why this feature?One rating can hide several source systems. This prototype shows the evidence trail without asking users to understand the data pipeline.

Sample Clinic — Gangnam

Rhinoplasty · revision rhinoplasty · Gangnam-gu, Seoul

4.5★★★★★Published rating

Review integrity check

From scattered reviews to a traceable signal

This prototype runs a transparent first-pass audit over an embedded multi-source sample. Deterministic checks narrow the records an AI translator or semantic matcher must inspect, while uncertain matches stay visible for editorial review.

Ready to inspect the embedded multi-source sample.
  1. 1
    Normalize identities

    Canonicalize source, author and review text while retaining every original record.

    —
  2. 2
    Group likely duplicates

    Compare normalized author fingerprints and token similarity. Borderline matches stay reviewable.

    —
  3. 3
    Check translation coverage

    Preserve Korean originals and flag records that lack an English translation for a later model-assisted pass.

    —
  4. 4
    Attach verification evidence

    Separate clinic, surgeon and procedure verification so one badge never implies all three.

    3 checks
✓ Clinic identity matched ✓ Surgeon credential source linked ○ Procedure evidence pending