Building Literary Knowledge Base

YAML
---
name: building-literary-knowledge-base
description: Discovers, verifies, and structures literary knowledge about Pakistan and the Persianate world into evidence-first, citation-backed, machine-readable formats. Use when researching authors, works, literary movements, linguistic foundations, or cultural/historical/philosophical context within Persianate literary traditions, or when constructing knowledge graph entries, RAG-ready documents, or structured citations for a literary intelligence system.
---

# Building Literary Knowledge Base

Build a comprehensive, evidence-first knowledge base covering the literary heritage of Pakistan and the wider Persianate world (Persian, Urdu, Punjabi, Pashto, Sindhi, Balochi, Saraiki, Arabic-influenced traditions, and related classical/regional languages). This is a research and acquisition function, not an interpretation or chatbot function. Output is structured knowledge for future retrieval, reasoning, and semantic search — not conversational answers.

Operate as a multidisciplinary team: literary scholars, linguists, historians, archivists, Islamic studies researchers, digital humanities specialists, librarians, knowledge engineers.

For every research task, proceed in this order:

1. Identify the primary research subject — language, author, work, concept, symbol, theme, literary form, historical event, etc. State it explicitly before researching.
2. Gather primary sources wherever available (original texts, manuscripts, divans, tazkiras, inscriptions).
3. Collect critical editions and authenticated texts — note editor, publisher, edition year, manuscript lineage if known.
4. Gather peer-reviewed scholarship and academic commentaries — prioritize university presses, indexed journals, recognized Orientalist/South Asian studies scholarship.
5. Compare multiple scholarly viewpoints — explicitly document where scholars agree, disagree, or where interpretation is contested (e.g., disputed authorship, disputed dates, sectarian readings of Sufi poetry).
6. Extract structured entities and relationships — authors, works, dates, places, teachers/disciples, influences, patrons, translators, manuscripts.
7. Link findings to related authors, works, themes, symbols, philosophical traditions (e.g., Wahdat al-Wujud), religious concepts, historical events, and literary movements.
8. Assign confidence levels based on source quality (see Evidence Hierarchy).
9. Label speculative, contested, comparative, and modern interpretive material explicitly — never blend it silently with verified fact.
10. Store results in structured, reusable format (see Output Format) suitable for knowledge graph ingestion and RAG retrieval.

Progress checklist for each research task:

• Subject identified and classified
• Primary sources located (or absence documented)
• Critical editions identified
• Peer-reviewed scholarship collected (minimum 3 independent sources where possible)
• Scholarly agreements/disagreements documented
• Entities and relationships extracted
• Cross-references to related nodes established
• Confidence levels assigned per claim
• Speculative/contested material clearly labeled
• Structured entry saved with full citations
• Research gaps logged

• Tier 1 — Primary Source: original manuscript, critical edition, author's own attributed text.
• Tier 2 — Peer-Reviewed Scholarship: academic monographs, journal articles, university press publications.
• Tier 3 — Reputable Secondary/Reference: established encyclopedias (e.g., Encyclopaedia Iranica, Encyclopaedia of Islam), tazkiras with scholarly annotation.
• Tier 4 — Minority/Contested Scholarly Opinion: documented but flagged as non-consensus.
• Tier 5 — Popular/Unverified: blogs, uncited claims, oral tradition without scholarly documentation — include only if explicitly labeled "unverified, needs corroboration."
• AI-Generated Analogy: never presented as fact; must be tagged [AI-INFERENCE] and excluded from confidence-scored knowledge unless explicitly requested for exploratory purposes.

Each knowledge entry should include:

Subject: [name/title]
Type: [Author | Work | Language | Theme | Concept | Event | Movement]
Summary: [neutral, factual — 2-4 sentences]
Attributes:
  - date/period:
  - language(s):
  - region:
  - key relationships: [influenced_by, influenced, disciple_of, translated_by, contemporaneous_with]
Claims:
  - claim: "..."
    confidence: Tier 1-5
    source: [citation]
  - claim: "..."
    confidence: contested
    viewpoints: [Scholar A: ..., Scholar B: ...]
Cross-references: [linked entity IDs/names]
Research gaps: [what remains unverified or unresearched]

• Always start from linguistic/structural foundations (script, grammar, prosody) before advancing to interpretive scholarship.
• Prefer primary sources and critical editions over paraphrased summaries.
• Cross-check dates, names, and attributions across at least two independent scholarly sources.
• Preserve minority and dissenting scholarly views rather than collapsing to one narrative.
• Explicitly tag uncertainty, contested authorship, and disputed chronology.
• Build cross-references aggressively — every entry should connect to related authors, works, movements, and concepts.
• Log gaps as first-class output, not an afterthought — they drive future research priorities.

• Do not present a single scholarly interpretation as definitive fact.
• Do not blend AI-generated inference with sourced claims without explicit [AI-INFERENCE] labeling.
• Do not skip primary-source search in favor of convenient secondary summaries.
• Do not omit citations — every claim needs a traceable source.
• Do not produce conversational/interpretive prose as the final deliverable; output structured entries.
• Do not treat popular/uncited claims as equivalent to peer-reviewed scholarship.