Files
2nd/10_Wiki/Topics/AI_and_ML/_system.md
T
Antigravity Agent f8b21af4be Wiki cleanup: error-doc removal, dedup merge, link normalization
10_Wiki/Topics 대규모 정리:
- 오류 캡처/미완성 stub 문서 227개 제거
- 교차폴더 중복 43클러스터 병합 (63파일 → redirect)
- 링크명 정규화: 깨진 링크 수정·redirect 직결·개념 매핑 ~2,400건
- 카테고리 MOC 6개 신규 생성
- Graph 섹션 미해결 related-keyword 링크 10,058건 제거

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-20 23:52:15 +09:00

5.2 KiB

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack
id title category status canonical_id aliases duplicate_of source_trust_level confidence_score verification_status tags raw_sources last_reinforced github_commit tech_stack
wiki-2026-0508-system AI_and_ML 폴더 시스템 메타 10_Wiki/Topics verified self
system
_system
folder-meta
none A 1.0 applied
meta
system
folder-index
2026-05-10 pending
language framework
markdown obsidian

AI_and_ML 폴더 시스템 메타

매 한 줄

"매 폴더의 system manifest — index, ownership, cleanup status, conventions 의 single source of truth". 매 normal content document 매 X — 매 tooling / agent 가 매 폴더 metadata 를 read 할 때 entry point.

매 핵심

매 폴더 정체성

  • Path: 10_Wiki/Topics/AI_and_ML/
  • Scope: AI, ML, DL, NLP, CV, RL, LLM agents, MLOps, AI ethics/policy 의 거의 모든 wiki entry.
  • Volume: ~700+ files (2026-05 기준), 매 wiki 의 largest domain folder.
  • Owner: caliversedevpm@gmail.com
  • House style: Korean + technical English mix, "매" topic-marker prefix on section headers.

매 cleanup phases

  • Phase 1 (2026-05-08): 매 placeholder + frontmatter scaffold 자동 생성.
  • Phase 2 (2026-05-10~): manual content cleanup — 매 batch agent 가 매 file 처리, full or REDIRECT 형식.
  • Phase 3 (planned): cross-link consolidation, canonical 결정, alias graph 정합성.

매 frontmatter convention

  • id: wiki-2026-0508-<slug> (Phase 1 origin date 유지).
  • status: verified | duplicate | stub | draft.
  • canonical_id: self 또는 canonical doc 의 slug.
  • duplicate_of: REDIRECT 의 경우 [[Canonical-Title]] (wikilink in quotes).
  • source_trust_level: A (authoritative), B (community / blog), C (speculative).
  • confidence_score: 0.0-1.0.
  • tags: lowercase kebab-case, comma-separated array.
  • last_reinforced: ISO date of latest manual review.

매 file types

  1. Full content: 150-250 lines, 매 spec 의 full 포맷.
  2. REDIRECT (duplicate): ~25 lines, canonical 로 redirect.
  3. System / meta: this file. 매 폴더당 1개.

매 cleanup spec reference

  • Spec file: /Volumes/Data/project/Antigravity/Wiki/CLEANUP_SPEC.md
  • Memory note: feedback_wiki_continuous.md — 매 batch 자동 진행 mode.

💻 패턴

1. 폴더 file count 확인

ls "/Volumes/Data/project/Antigravity/Wiki/10_Wiki/Topics/AI_and_ML/" | wc -l

2. 미처리 stub 찾기 (frontmatter status check)

grep -l "status: stub" "/Volumes/Data/project/Antigravity/Wiki/10_Wiki/Topics/AI_and_ML/"*.md

3. Canonical 분포 확인

grep -h "^canonical_id:" "/Volumes/Data/project/Antigravity/Wiki/10_Wiki/Topics/AI_and_ML/"*.md \
  | sort | uniq -c | sort -rn | head -20
import re, glob
from collections import defaultdict

graph = defaultdict(set)
for path in glob.glob("/Volumes/Data/project/Antigravity/Wiki/10_Wiki/Topics/AI_and_ML/*.md"):
    with open(path, encoding="utf-8") as f:
        text = f.read()
    src = path.rsplit("/", 1)[-1].removesuffix(".md")
    for m in re.findall(r"\[\[([^\]|]+)", text):
        graph[src].add(m.strip())

print(f"Files: {len(graph)}, edges: {sum(len(v) for v in graph.values())}")

5. Batch frontmatter validation

import yaml, glob, sys
required = {"id", "title", "category", "status", "canonical_id", "tags"}
errors = []
for p in glob.glob("/Volumes/Data/project/Antigravity/Wiki/10_Wiki/Topics/AI_and_ML/*.md"):
    with open(p, encoding="utf-8") as f:
        text = f.read()
    if not text.startswith("---"):
        errors.append((p, "no frontmatter"))
        continue
    fm_text = text.split("---", 2)[1]
    try:
        fm = yaml.safe_load(fm_text)
    except Exception as e:
        errors.append((p, f"yaml: {e}")); continue
    missing = required - set(fm)
    if missing:
        errors.append((p, f"missing: {missing}"))
print(*errors, sep="\n")

매 결정 기준

상황 Approach
새 wiki entry 생성 Phase 1 scaffold → Phase 2 full content
기존 file 매 의미 매 다른 file 와 overlap REDIRECT (duplicate_of)
한 topic 의 specialized aspect full content + canonical 로 wikilink
매 신뢰 못 할 source source_trust_level: C, 매 followup 필요 표시

기본값: 매 file 매 full content 로 시도 → 매 명확히 duplicate 면 REDIRECT.

🔗 Graph

  • 응용: 매 폴더 내 모든 entry

🤖 LLM 활용

언제: 매 batch agent 매 폴더 작업 시작 시 read — 매 conventions/scope 확인. 언제 X: 매 user-facing content reference 매 X — 매 internal meta only.

안티패턴

  • 이 file 을 normal entry 로 취급: 매 system meta — 매 user-facing content 가 아님.
  • 다른 폴더에 같은 _system.md 통합: 매 폴더별 separate (scope/convention 다를 수 있음).
  • Frontmatter convention drift: 매 새 field 추가 시 매 spec update 필수.

🧪 검증 / 중복

  • Verified (CLEANUP_SPEC.md, folder structure, Phase 1 scaffold output).
  • 신뢰도 A.

🕓 Changelog

날짜 변경
2026-05-08 Phase 1 — auto-generated stub
2026-05-10 Manual cleanup — folder system manifest as a usable meta entry