Mastering K-Beauty Data: Implementing HSK 2025 Standards & Smart Classification

Key Takeaways (요약)

1. HSK 2025 최신 분류 체계 적용: 마스크팩, 필러 등 혼동하기 쉬운 품목의 HS 코드를 명확히 재정의하고 16개 카테고리로 세분화했습니다.
2. 풍부한 검색 키워드 매핑: '토너', '세럼' 등 실무에서 사용하는 다양한 용어를 HS 코드와 연동하여 검색 정확도를 높였습니다.
3. 동적 유의사항(Notice) 시스템: 데이터 해석의 오류를 방지하기 위해 선택한 카테고리에 맞는 안내 문구를 즉각적으로 표시하는 UI를 구현했습니다.

Why Precise Data Classification Matters in K-Beauty AI

In the world of international trade, data accuracy is paramount. For the K-Beauty Export Tracker, relying on generic HS Codes often leads to misleading insights. For instance, a "Sheet Mask"—one of Korea's biggest exports—is technically classified differently from a "Gel Pack," yet they are often lumped together in casual conversation. To provide actionable intelligence, our AI must distinguish these nuances with customs-level precision.

The Problem: "Is this a Mask or a Lotion?"

Under the strict Korea Customs Service (KCS) standards:

• Sheet/Patch Masks fall under 3307.90 (Other perfumery/toilet preparations).
• Gel/Paste Masks fall under 3304.99 (Basic Skincare), alongside lotions and creams.

Without a smart mapping engine, a user searching for "Mask Pack" might see data excluding half the market. We needed a system that bridges the gap between consumer terms and customs codes.

Solution: The Intelligent HSK Mapping System

We overhauled our classification logic to align with the HSK 2025 standards. The core of this update is a comprehensive Typescript constant that acts as the source of truth for the entire application.

1. Robust Data Structure

We defined a HSKItem interface that holds not just the code and label, but also rich metadata: specific keywords for search and a note field for contextual warnings.

// frontend/lib/constants/hsk.ts
 
export interface HSKItem {
    id: string;
    group: 'Basic' | 'Makeup' | 'Hair' | 'Cleansing' | 'Other';
    label: string;
    hsCode: string; // The primary key for API calls
    note?: string;   // Contextual warning for the user
    keywords: string[]; // Rich keywords for search optimization
}

2. Rich Keyword Mapping

We didn't just label it "Basic Skincare". We mapped it to every term a user might type—English or Korean.

// Example: Basic Skincare Mapping
{
    id: 'basic_set',
    group: 'Basic',
    label: "기초화장용 제품류 (Basic Skincare)",
    hsCode: "3304991000",
    note: "본 데이터는 기초화장용 제품류(3304.99.1000) 통계를 기반으로 합니다. 스킨·토너, 로션, 에센스...",
    keywords: [
        "토너", "스킨", "로션", "에멀전", "세럼", "에센스", "영양크림",
        "Toner", "Lotion", "Essence", "Serum", "Cream"
    ]
}

This approach allows our search bar to function like a semantic search engine. Typing "Serum" instantly routes the user to the correct 3304.99.1000 HS Code data.

Enhancing UX: Context-Aware Notices

Data is only useful if it's interpreted correctly. A common pitfall for analysts is assuming 3304.20 covers all eye makeup, forgetting that "Lip & Eye Removers" are often grouped there due to customs rules, not product function.

To address this, we implemented a Dynamic Notice System.

What is the most efficient way to prevent data misinterpretation?

Instead of a generic FAQ page, we placed the information exactly where the user needs it: right inside the dashboard, triggered by their selection.

// frontend/components/Dashboard.tsx
 
{/* Category Disclaimer Note */}
{selectedItem.note && (
  <div className="mb-6 p-4 bg-amber-50 dark:bg-amber-900/20 border border-amber-200 rounded-lg flex items-start gap-3">
    <span className="text-amber-500 mt-0.5">ⓘ</span>
    <p className="text-sm text-amber-800 dark:text-amber-200">
       {selectedItem.note}
    </p>
  </div>
)}

This simple yet effective logic checks if the currently selected mapping item has a note property. If it does, it renders a high-visibility alert box explaining exactly what is included (or excluded) in the current data view.

Impact on Discovery

This update significantly improves the reliability of our "Rising Stars" discovery feature. By clearly segregating items like Functional Cosmetics (3304.99.9000) comprising high-value fillers, investors can track premium market segments separately from mass-market skincare.

Comparison: Old vs. New Architecture

Why we prioritize "Data Transparency"

In building an AI agent for global business, trust is the currency. By exposing the raw definitions and caveats of our data through these specific notices, we empower our users—brand managers, investors, and marketers—to make decisions with eyes wide open.

This update effectively closes the gap between raw trade statistics and actionable business intelligence.

This project is part of our journey to expand K-Beauty globally using AI.