Output JSON Scheme

Overview

The Research → Seed pipeline discovers entry_url from source_domain through a 5-stage multi-step exploration and generates Output JSON.

Pipeline Stages

[1] Well-known RSS Feeds (11개 표준 경로)
   ↓ (없으면)
[2] Sitemap.xml 분석 (뉴스/미디어 URL 추출)
   ↓ (없으면)
[3] Robots.txt 힌트 (Sitemap/Allow 경로)
   ↓ (없으면)
[4] HTML 패턴 탐색 (RSS 링크, 네비게이션)
   ↓ (없으면)
[5] 수동 검토 필요 (자동화 한계)

Output JSON Structure

{
 "meta": {
   "generated_at": "2026-01-24T14:23:45.777Z",
   "total_seeds": 6,
   "pipeline_version": "2.0",
   "generator": "newsfork-seed-discovery"
 },
 "seeds": [
   {
     "id": "sg-mom-001",
     "source_domain": "mom.gov.sg",
     "entry_url": "https://www.mom.gov.sg/newsroom.xml",
     "nature": "news",
     "fetch": {
       "type": "html",
       "selectors": {
         "container": "main, .content, .news-content"
       }
     },
     "url_patterns": ["..."],
     "schedule": "*/30 * * * *",
     "priority": 9,
     "trust_tier": "tier1",
     "status": "draft",
     "metadata": {
       "discovery_confidence": 0.85,
       "nature_detection": {
         "method": "url_pattern",
         "confidence": 0.9
       }
     }
   }
 ]
}

Content Nature Automatic Classification

URL Pattern	Nature	Confidence
/newsroom/	news	90%
/press/	press_release	85%
/faq/	faq	80%
/guide/	guide	80%

Research vs Seed Role Separation

• Research: Discover “where to look” (URL list only)
• Seed: Contract “how to retrieve” (nature identification + collection strategy)