WFGY/ProblemMap/GlobalFixMap/Language/multilingual_guide.md

Multilingual Guide — Guardrails and Fix Patterns

🧭 Quick Return to Map

You are in a sub-page of Language.
To reorient, go back here:

• Language — multilingual processing and semantic alignment
• WFGY Global Fix Map — main Emergency Room, 300+ structured fixes
• WFGY Problem Map 1.0 — 16 reproducible failure modes

Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.

A compact field guide to stabilize multilingual RAG across CJK, RTL, mixed scripts, and locale drift. Use this page to check symptoms, apply structural fixes, and verify with measurable targets.

---

Open these first

• Visual recovery map: RAG Architecture & Recovery
• Tokenization vs semantics: Embedding ≠ Semantic
• OCR parsing in multilingual docs: OCR Parsing Checklist
• Chunk joins across scripts: Semantic Chunking Checklist
• Why this snippet: Retrieval Traceability
• Snippet schema enforcement: Data Contracts

---

Core acceptance

• ΔS(question, retrieved) ≤ 0.45 across at least 2 languages
• Coverage ≥ 0.70 for the target section in each language
• λ remains convergent across three paraphrases in mixed scripts
• E_resonance stays flat for long bilingual/RTL runs

---

Common multilingual failure modes

Symptom	Likely cause	Open this
Retrieval drops snippets when query is in Chinese or Japanese	Tokenizer mismatch (no whitespace segmentation)	tokenizer_mismatch.md
Citations collapse when Arabic/Hebrew text mixes with English	Script directionality conflict	script_mixing.md
High similarity but meaning flips across locale	Locale analyzer mismatch (stemming / stopwords)	locale_drift.md
HyDE/BM25 retrieval different per language	Query expansion language bias	hyde_multilingual.md

---

Fix in 60 seconds

1. Probe with ΔS Run the same question in English and one target language. If ΔS differs by >0.15, suspect tokenization or analyzer mismatch.

2. Apply λ_observe Paraphrase the query three ways in the non-English language. If λ diverges, enforce schema lock and re-index with language-specific analyzers.

3. Structural repair

   • Tokenizer mismatch → tokenizer_mismatch.md
   • Script mixing issues → script_mixing.md
   • Locale drift → locale_drift.md
   • HyDE instability → hyde_multilingual.md

---

Diagnostic checklist

• Tokenizer: verify segmentation strategy (whitespace vs character-level)
• Analyzer: confirm stemming and stopword lists match query language
• Scripts: normalize Unicode, check RTL/LTR flags
• Locale drift: run same snippet under two locales, compare ΔS
• Hybrid retriever: ensure rerankers operate on normalized embeddings

---

🔗 Quick-Start Downloads (60 sec)

Tool	Link	3-Step Setup
WFGY 1.0 PDF	Engine Paper	1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>”
TXT OS (plain-text OS)	TXTOS.txt	1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly

---

Explore More

Layer	Page	What it’s for
Proof	WFGY Recognition Map	External citations, integrations, and ecosystem proof
Engine	WFGY 1.0	Original PDF based tension engine
Engine	WFGY 2.0	Production tension kernel and math engine for RAG and agents
Engine	WFGY 3.0	TXT based Singularity tension engine, 131 S class set
Map	Problem Map 1.0	Flagship 16 problem RAG failure checklist and fix map
Map	Problem Map 2.0	RAG focused recovery pipeline
Map	Problem Map 3.0	Global Debug Card, image as a debug protocol layer
Map	Semantic Clinic	Symptom to family to exact fix
Map	Grandma’s Clinic	Plain language stories mapped to Problem Map 1.0
Onboarding	Starter Village	Guided tour for newcomers
App	TXT OS	TXT semantic OS, fast boot
App	Blah Blah Blah	Abstract and paradox Q and A built on TXT OS
App	Blur Blur Blur	Text to image with semantic control
App	Blow Blow Blow	Reasoning game engine and memory demo

If this repository helped, starring it improves discovery so more builders can find the docs and tools.