Skip to content

Pāṇini File System

Metalinguistic information processing and storage platform. A universal semantic model of 34 atoms, validated across 14 languages and ~8 million words.

🔬 The Main Model — 34 Universal Atoms

The core of PaniniFS is a minimal semantic vocabulary of 34 atoms distributed across 4 ontological categories and 6 layers of abstraction. These atoms encode the essentials of human concepts in any language.

The 4 Ontological Categories

Canonical identifiers

Atom names are uppercase canonical identifiers, language-independent. Most are French-named; emotional axes (SEEKING, FEAR, CARE…) use Panksepp's English nomenclature. English meanings are shown in parentheses.

Category Sanskrit Atoms — canonical identifiers (with English meaning)
PROCESS kriyā MOUVEMENT (movement), COGNITION, COMMUNICATION, CRÉATION (creation), SEEKING, FEAR, CARE, GRIEF…
RELATION sambandha RELATION, STRUCTURE, INVARIANCE, DOMINATION, ORDRE (order)…
QUALITY guṇa BON (good), GRAND (big), VRAI (true), INTENSE, ANCIEN (old), MESURE (measure), PERCEPTION…
ENTITY dravya CHOSE (thing), AGENT, CORPS (body), LIEU (place), MATIÈRE (matter), EXISTENCE…

→ Complete table of the 34 atoms

The 7 Informational Dhātu Operators

Additionally, 7 dhātu operators encode information flows at a high level: COMM · ITER · TRANS · DECIDE · LOCATE · GROUP · SEQ

→ Dhātu Framework

📊 Validated Results — February 2026

7/7 European languages ≥ 90% lexical coverage

Gutenberg original corpus (11 files, modern texts):

Language Coverage
English 94.4%
Esperanto 93.2%
German 91.1%
Finnish 90.6%
Spanish 90.1%
French 90.1%
Italian 90.1%

Major Multilingual Breakthroughs

Language Before After Gain Key technique
🇯🇵 Japanese 18.8% 74.1% +55pp Kanji-only tokenization + furigana stripping
🇨🇳 Chinese 33.8% 73.9% +40pp OpenCC traditional→simplified
🇷🇺 Russian 16.5% 56.3% +40pp Snowball stemmer + pre-1918 normalization
🇳🇱 Dutch 28.4% 55.9% +28pp Pre-1947 spelling normalization

Global coverage

76.8% across 62 Gutenberg texts + 973 Wikipedia articles (~8M words, 14 languages). 34/34 atoms present = 100% on the multilingual Wikipedia corpus.

→ Detailed coverage results · What's new

🔑 Key Discovery

The semantic atom is independent of writing system. Japanese kanji share the same characters as Chinese hanzi — coverage gained for Chinese directly benefits Japanese. This confirms that the 34 dhātu atoms are genuine conceptual universals, transcending writing systems.

🦀 PaniniWeb — Decentralized Architecture (Rust v0.1)

  • 4 crates workspace: panini-core, panini-net, panini-api, panini-cli
  • 71 tests (58 core + 11 net + 2 doc)
  • P2P network: libp2p with mDNS, Gossipsub, Kademlia
  • panini:// URI scheme — decentralized semantic web

🌍 Social Vision and Ethics

Society before tech. Goal: make information truly useful, accessible and traceable to everyone.

  • Inclusion and accessibility by default
  • Attribution and idea provenance (collective memory)
  • Open governance, aligned with the Montreal Declaration

→ Social vision · References

Quick Navigation

Section Description
Research Overview of research axes
Universal Atoms (34) Full table with NSM, Jackendoff, dhātu
Coverage Results Detailed metrics by language and corpus
Dhātu Framework The 7 operators + 34 atoms
Progress & roadmap Project status and roadmap
Book Complete documentation

Find Me