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>
5.3 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-advanced-search-operators | Advanced Search Operators | 10_Wiki/Topics | verified | self |
|
none | A | 0.9 | applied |
|
2026-05-10 | pending |
|
Advanced Search Operators
매 한 줄
"매 search operator는 매 검색의 inverse index에 대한 직접 명령". 매 1990년대 boolean 검색에서 출발해 매 2026 LLM-augmented search (Perplexity, Claude search, GPT-5 browse) 시대에도 매 underlying engine은 여전히 operator-driven, 매 power user 의 productivity multiplier.
매 핵심
매 Operator 분류
- Boolean:
AND,OR,NOT(or-) — 매 logical combination. - Phrase:
"exact phrase"— 매 token sequence 강제. - Field-restrict:
site:,intitle:,inurl:,filetype:,intext:— 매 specific field 검색. - Range / numeric:
2020..2025,before:2026-01-01,after:2025-06-01. - Wildcard:
*— 매 missing word. - Cache / archive:
cache:,web.archive.org/web/*/url.
매 엔진별 차이
- Google: 매 strict,
site:,filetype:,intitle:,before:,after:지원. - Bing:
site:,language:,loc:,feed:. - DuckDuckGo:
!bang(!w wikipedia,!gh github). - GitHub:
repo:,path:,language:,extension:,is:issue is:open. - Perplexity / Claude: 매 natural language 도 OK 지만 매 explicit operator 가 더 reliable.
매 응용
- OSINT: 매 leaked credential 검색 (
"@company.com" filetype:txt site:pastebin.com). - Research:
site:arxiv.org "diffusion transformer" after:2025-01-01. - Debugging:
site:stackoverflow.com [exact error message]. - Competitive intel:
site:competitor.com filetype:pdf intitle:"roadmap".
💻 패턴
Pattern 1: Site-restricted academic search
site:arxiv.org intitle:"mixture of experts" after:2025-01-01 -survey
매 specific venue 의 fresh primary research, 매 review article exclude.
Pattern 2: GitHub code archaeology
repo:anthropics/claude-code path:**/*.ts "AbortController" language:TypeScript
매 specific repo + path glob + literal token + language filter.
Pattern 3: Wayback Machine snapshot
https://web.archive.org/web/2024*/openai.com/pricing
매 historical pricing change track (매 2024 모든 snapshot).
Pattern 4: Filetype hunt
"annual report" filetype:pdf site:tesla.com
매 specific document format 의 corporate disclosure.
Pattern 5: Negative filtering
"react server components" -tutorial -beginner -"how to"
매 advanced content only, 매 entry-level material exclude.
Pattern 6: Boolean composition
("vector database" OR "vector store") AND (pgvector OR qdrant) -benchmark
매 synonym expansion + scope narrowing.
Pattern 7: Numeric range
"GPU memory" 40..80GB site:nvidia.com
매 spec range 검색.
Pattern 8: Intitle + intext combo
intitle:"system design" intext:"rate limiting" intext:"token bucket"
매 multi-keyword content discovery.
Pattern 9: Stack Overflow targeted
site:stackoverflow.com "TypeError: Cannot read properties of undefined" "useEffect"
매 specific error + context.
Pattern 10: DDG bang chaining
!gh microsoft/vscode editor.contribution
매 instant redirect to GitHub 매 search.
매 결정 기준
| 상황 | Approach |
|---|---|
| 매 fresh research | after: + site:arxiv.org / site:openreview.net |
| 매 specific error | exact "message" + site:stackoverflow.com |
| 매 corporate intel | filetype:pdf site:company.com |
| 매 code search | GitHub repo: + path: + language: |
| 매 historical | Wayback web.archive.org/web/<year>*/url |
| 매 ambiguous topic | Boolean OR + negative filter |
기본값: 매 quoted phrase + site: + after: 조합 — 매 noise 의 80% 제거.
🔗 Graph
- 부모: Search · Information Retrieval
- 응용: Codebase_Onboarding · Research-Methodology
- Adjacent: Pyramid Principle · Knowledge synthesis
🤖 LLM 활용
언제: 매 LLM agent 가 web search tool 호출 시 — 매 explicit operator 로 query 작성하면 매 retrieval precision 급상승. 매 Claude / GPT-5 의 browse tool 도 매 underlying Google/Bing 사용 → operator pass-through. 언제 X: 매 conceptual / synthesis question — 매 LLM 이 직접 reasoning. 매 operator 는 매 specific document/fact retrieval 용.
❌ 안티패턴
- 너무 많은 operator: 매 5개 이상 stacking 매 zero result. 매 progressive narrowing 의.
- Quoted long phrase:
"the quick brown fox jumps over"— 매 too specific. 매 3-5 word key phrase 가 sweet spot. site:over-restriction: 매 single domain 만 보면 매 broader context 손실.- Stale cache reliance:
cache:는 Google 에서 2024 deprecated — 매 web.archive.org 사용.
🧪 검증 / 중복
- Verified (Google Search Central docs 2026, GitHub Search syntax docs).
- 신뢰도 A.
🕓 Changelog
| 날짜 | 변경 |
|---|---|
| 2026-05-08 | Phase 1 |
| 2026-05-10 | Manual cleanup — operator taxonomy + 10 patterns + LLM browse 매 통합 |