101 lines
4.9 KiB
Markdown
101 lines
4.9 KiB
Markdown
# PHASE INDEX - ontology_platform engine-respect roadmap
|
||
|
||
?묒꽦?? 2026-05-19
|
||
|
||
踰붿쐞: `ontology_platform` ?꾩슜. `crawler_platform`? ?대쾲 ?묒뾽 踰붿쐞?먯꽌 ?쒖쇅?쒕떎.
|
||
|
||
湲곗? 臾몄꽌:
|
||
- `ontology_platform/docs/?듯빀?ㅺ퀎??md`
|
||
- `ontology_platform/README.md`
|
||
- `ontology_platform/docs/phases/PHASE0_ACCEPTANCE_GATE.md`
|
||
- `ontology_platform/docs/phases/PHASE1_NEXT_STEPS.md`
|
||
- `ontology_platform/docs/phases/PHASE1_ACCEPTANCE_GATE.md`
|
||
- `ontology_platform/docs/phases/PHASE2_ACCEPTANCE_GATE.md`
|
||
|
||
?듭떖 ?먯튃:
|
||
- OntoCast??Base ?붿쭊?쇰줈 議댁쨷?쒕떎.
|
||
- vendored OntoCast 肄붿뼱???듯빀?ㅺ퀎?쒓? ?덉슜??踰붿쐞 ?몄뿉???섏젙?섏? ?딅뒗??
|
||
- Trafilatura, Crawl4AI, Guardrails, Neo4j GraphRAG??吏곸젒 ?ш뎄?꾪븯吏 ?딄퀬 ?뉗? adapter/facade濡?媛먯떬??
|
||
- Firecrawl, OpenDeepResearcher 肄붾뱶???ы븿?섏? ?딅뒗??
|
||
- Acceptance Gate瑜??듦낵?섍린 ???ㅼ쓬 ?듯빀?쇰줈 ?섏뼱媛吏 ?딅뒗??
|
||
|
||
---
|
||
|
||
PHASE 0. ?붿쭊 寃쎄퀎 媛먯궗 諛?Phase Gate 蹂듦뎄
|
||
FILE: ./26_05_19_engine_respect_plan/phase_00_001_engine_boundary_gate.md
|
||
|
||
1) ?꾩옱 `ont_platform` 紐⑤뱢??Base/Adapter/Draft/Excluded 梨낆엫?쇰줈 遺꾨쪟 [?꾨즺]
|
||
2) Phase 0?먯꽌 誘몃옒 Phase ?섏〈?깆씠 import?섏뼱 ???쒖옉??源⑥? ?딅룄濡?寃뚯씠???뺣━ [?꾨즺]
|
||
3) Phase 0 unit/integration 寃利??덉감 怨좎젙 [?꾨즺]
|
||
4) `PHASE0_ACCEPTANCE_GATE.md` 媛깆떊 湲곗? ?뺣━ [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 1. Trafilatura 湲곕컲 URL/HTML ?낅젰 ?뺣젹
|
||
FILE: ./26_05_19_engine_respect_plan/phase_01_001_trafilatura_ingestion.md
|
||
|
||
1) `web_extractor.py`瑜?Trafilatura adapter 梨낆엫?쇰줈 ?뺣━ [?꾨즺]
|
||
2) `SourceDocument`, `EvidenceSpan`, Content metadata ???寃쎄퀎 ?곌껐 [?꾨즺]
|
||
3) `/process/url` ?먮뒗 ?숇벑??URL ?낅젰 API ?ㅺ퀎 [?꾨즺]
|
||
4) ?쒓뎅??URL/HTML fixture 湲곕컲 異붿텧 ?뚯뒪?몄? dedup 湲곗? ?묒꽦 [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 2. Candidate Storage 諛?Review 梨낆엫 寃쎄퀎
|
||
FILE: ./26_05_19_engine_respect_plan/phase_02_001_candidate_review_boundary.md
|
||
|
||
1) `storage/models.py`???꾨낫 紐⑤뜽???뺤떇 Review Queue 怨꾩빟?쇰줈 ?뺤젙 [?꾨즺]
|
||
2) OntoCast 寃곌낵? lightweight extraction 寃곌낵?????寃쎈줈 遺꾨━ [?꾨즺]
|
||
3) ?뱀씤/諛섎젮/?먮룞?뱀씤 ?곹깭 ?꾩씠 洹쒖튃 ?뺤쓽 [?꾨즺]
|
||
4) evidence ?녿뒗 ?꾨낫媛 ?뺤젙 graph濡??ㅼ뼱媛吏 紐삵븯寃?李⑤떒 [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 3. Crawl4AI ?섏쭛 怨꾩링 諛?Job Orchestration
|
||
FILE: ./26_05_19_engine_respect_plan/phase_03_001_crawl4ai_acquisition_jobs.md
|
||
|
||
1) `crawl4ai_adapter.py`瑜??숈쟻/????섏쭛 adapter濡??쒗븳 [?꾨즺]
|
||
2) crawler profile, robots policy, cache policy瑜??ㅼ젙 湲곕컲?쇰줈 遺꾨━ [?꾨즺]
|
||
3) Job ?곹깭 紐⑤뜽怨?progress API/WebSocket 寃쎄퀎 ?뺣━ [?꾨즺]
|
||
4) Trafilatura ?꾩쿂由ъ? SourceDocument ??μ쑝濡??곌껐 [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 4. Guardrails Validation Gate
|
||
FILE: ./26_05_19_engine_respect_plan/phase_04_001_guardrails_validation_gate.md
|
||
|
||
1) `core/validation`??Pydantic lightweight? Guardrails facade濡?遺꾨━ [?꾨즺]
|
||
2) OntoCast LLM 異쒕젰 ?섑븨 吏?먯쓣 vendored ?섏젙 ?놁씠 ?곗꽑 ?ㅺ퀎 [?꾨즺]
|
||
3) schema violation, endpoint missing, confidence range ?뚯뒪???묒꽦 [?꾨즺]
|
||
4) Guard ?ㅽ뙣 寃곌낵瑜?candidate/review issue濡????[?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 5. Neo4j Projection 諛?GraphRAG 寃??FILE: ./26_05_19_engine_respect_plan/phase_05_001_neo4j_projection_graphrag.md
|
||
|
||
1) RDF/Fuseki瑜?canonical store, Neo4j瑜?projection/search store濡?怨좎젙 [?꾨즺]
|
||
2) `core/graph` 湲곗〈 紐⑤뱢??projection/search adapter 梨낆엫?쇰줈 ?щ텇瑜?[?꾨즺]
|
||
3) read-only Text2Cypher? vector/hybrid retriever API ?ㅺ퀎 [?꾨즺]
|
||
4) provenance媛 search result源뚯? ?댁뼱吏??寃利?湲곗? ?묒꽦 [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 6. Maintenance Loop 諛??댁쁺 湲곕뒫 ?뺣━
|
||
FILE: ./26_05_19_engine_respect_plan/phase_06_001_maintenance_loop_operations.md
|
||
|
||
1) Knowledge Agent??肄붾뱶媛 ?꾨땲???꾨＼?꾪듃/?뚰겕?뚮줈???⑦꽩留?李⑥슜 [?꾨즺]
|
||
2) Analyst/Researcher/Curator/Auditor/Fixer/Advisor 梨낆엫 ?뺤쓽 [?꾨즺]
|
||
3) `auth`, `audit`, `billing`, `realtime` 珥덉븞 紐⑤뱢???댁쁺 寃쎄퀎 ?뺣━ [?꾨즺]
|
||
4) destructive fix???щ엺 ?뱀씤 寃뚯씠?몃? 諛섎뱶???듦낵?섎룄濡??ㅺ퀎 [?꾨즺]
|
||
|
||
---
|
||
|
||
PHASE 7. Hybrid Rule + LLM Extraction
|
||
FILE: ./26_05_19_engine_respect_plan/phase_07_001_hybrid_rule_llm_extraction.md
|
||
|
||
1) rule baseline, LLM extraction, fallback, validation, Review UI 흐름을 기준선으로 고정 [신규]
|
||
2) `rule_only`, `llm_only`, `hybrid`, `compare` extraction mode 계약 정의 [신규]
|
||
3) product backend에 명시적 HybridExtractor와 rule/LLM agreement metadata 추가 [신규]
|
||
4) confidence breakdown에 rule agreement와 conflict/review 정책 반영 [신규]
|
||
5) Crawl/Research UI에서 mode/provider/model/base URL 선택 지원 [신규]
|