Skip to content

schema.md

Latest commit

 

History

History
74 lines (65 loc) · 2.03 KB

schema.md

File metadata and controls

74 lines (65 loc) · 2.03 KB
Raw

Enox $50 Experiment — Schema

Edge format (one per JSONL line)

{
  "_type": "edge",
  "from": "arxiv:XXXX.XXXXX",
  "rel": "extends",
  "to": "arxiv:YYYY.YYYYY",
  "confidence": 0.25,
  "condition": "sparse graphs only",
  "source": "arxiv:XXXX.XXXXX",
  "extracted": "2026-03-14",
  "status": "extracted",
  "note": "Brief human-readable explanation"
}

Node format (one per JSONL line)

{
  "_type": "node",
  "id": "arxiv:XXXX.XXXXX",
  "node_type": "paper",
  "title": "Paper Title",
  "authors": ["Author1", "Author2"],
  "year": 2024,
  "arxiv_id": "XXXX.XXXXX",
  "doi": "10.XXXX/...",
  "domain": "cs.LG",
  "citations_approx": 1500,
  "abstract_short": "One-line summary"
}

Concept nodes:

{
  "_type": "node",
  "id": "enox:concept/graph_neural_network",
  "node_type": "concept",
  "label": "Graph Neural Network",
  "domain": "cs.LG",
  "aliases": ["GNN"]
}

Relation archetypes

Relation Domain What it captures
implements CS Paper implements algorithm from another
extends CS Paper extends/improves upon method
outperforms CS Method beats another (with conditions)
fails_on CS Method fails under conditions
requires CS Method requires technique/assumption
introduces CS Paper introduces a new concept/method
supports Science Provides evidence for a claim
refutes Science Contradicts findings
is_based_on CS Theoretical foundation
applies_to CS Method applicable to domain/problem
isomorphic_to Meta Structural similarity across domains
supersedes CS Newer method replaces older
surveys CS Paper reviews a field
enables CS One technique enables another
formalizes CS/Math Paper formalizes an informal concept

Confidence levels (all LLM-extracted = 0.1-0.3)

  • 0.1 = weak inference from general knowledge
  • 0.2 = moderate confidence, based on well-known relationships
  • 0.3 = high confidence for LLM extraction (e.g., explicit in abstract)