# Maqāyīs (مَقاييس) — Open Arabic Computational Philology & Zero-LLM NLP

> Unified open-source research organization, scholarly monograph series, algebraic lexical-geometric atlases, Zero-LLM retrieval engines, governed scientific machine translation, and interactive academy for Arabic computational linguistics.
> Author & Principal Investigator: Salah Abdo (<Salah.Abdo.Tech@gmail.com>)
> Canonical Portal: https://maqayis.org
> GitHub Organization: https://github.com/maqayis
> Licensing: Apache-2.0 (all software, SDKs, portal, edge worker) | CC BY 4.0 (Book v2.0, public lexical atlases, AWN4 static index, Dīwān ledger, public Manhaj curriculum)

## Core Pillars & Public Portal Routes

- [Interactive Book v2.0 & Monographs (`#/book`)](https://maqayis.org/#/book): *Handasat al-Lisān al-ʿArabī* (*هندسة اللسان العربي — الأسس الرياضية والحاسوبية للغة العربية*, v2.0), 10 companion monographs (7 Arabic + 3 English preprints), 140-concept formal glossary, and 75-claim reproducible empirical verification ledger.
- [Unified Lexicon & Root Geometry Hub (`#/lexicon`)](https://maqayis.org/#/lexicon):
  - **Al-Jadwal (`28³` Periodic Table of Arabic Roots)**: 11,249 roots across 6 derivational strata (`S0–S5`), 716 bilateral `[C₁C₂•]` subroot genera, 7 Qurʾānic *Furūq* minimal-pair clusters, 83 gold safe *Muhmal* cells, and `28×28` biomechanical/orthographic OCP matrices.
  - **Iqtirān (`66`-Operator Morpho-Syntactic Algebra)**: Closed inventory of 31 morphological operators (`B01–B31`) × 35 relational/attachment operators (`R01–R35` = 1,085 phase-coupling cells) with verified classical *shawāhid*.
  - **Dīwān al-Iṣṭilāḥ**: 161 governed scientific terminology concepts (`cs.CL`, `cs.IR`) with `U → F1 → F2` discourse folding, 8-form paradigms, 6-gate phonotactic/morphological verification, and classical OpenITI bibliographic witnesses.
  - **Arabic WordNet v4 (`AWN4`)**: 120,630 synsets, 136,041 lexical entries, 184,238 senses, and 297,150 semantic relations served via deterministic `fnv1a_32(synset_id) & 0x7f` static shards.
- [Zero-LLM Tarkīb + Tayf Live Lab (`#/lab`)](https://maqayis.org/#/lab): Client-side `<5 ms` geometric Arabic retrieval arena, 5-Law `32`-bitmask ablation switchboard (`90.0% P@1`, `0.9262 MRR` with zero neural parameters vs `560M` `multilingual-e5-large`), 20 zero-giveaway trap queries, live custom Arabic query scorer, and live Arabic OCP phonotactic root tester.
- [Governed Scientific MT X-Ray Reader & BYOK Studio (`#/mt`)](https://maqayis.org/#/mt): Sentence-synchronized EN↔AR showcase papers (*ColBERT*, *LoRA*, *Multilingual-E5*) comparing `C3 Governed` (`100%` term adherence, `0` defects) vs `C0 Baseline`, interactive `U / F1 / F2` X-Ray term badges, and 100% client-side `BYOK` browser translator (`sessionStorage` only; zero server LLM custody).
- [Manhaj Interactive Academy (`#/academy` & `/manhaj/`)](https://maqayis.org/manhaj/): Structured bilingual tracks in Arabic computational philology, sub-lexical algebra, information theory, and governed NLP.
- [Community Governance, Open Benchmarks & Institutional Gateway (`#/community`)](https://maqayis.org/#/community): Moderated `Dīwān` term proposals, 5-suite `Maqāyīs Open Benchmarks` leaderboard, opt-in double-scrubbed `Misbār` research telemetry with 1-click cryptographic right-to-erasure, and institutional research collaboration (`research@maqayis.org`).

## Open-Source Repositories (`https://github.com/maqayis`)

- `maqayis/maqayis`: Umbrella specification, unified web portal (`portal/`), zero-cost Cloudflare Edge Observatory (`edge-worker/`), and release audit gates (`Apache-2.0`).
- `maqayis/tarkib` (`v2.0.0`): Computational philology, sub-lexical algebraic semantics, `Al-Jadwal` periodic root atlas, undotted *rasm* Viterbi solver, and `Iqtirān` 66-atom morpho-syntactic engine (`Apache-2.0` code / `CC BY 4.0` data & book).
- `maqayis/tayf` (`v1.0.0`): Standalone Zero-LLM Grassmannian, ontological, and metaphor-aware (`T_majāz`) Arabic RAG & retrieval benchmark arena (`Apache-2.0`).
- `maqayis/diwan` (`v1.0.0`): Cross-document Arabic scientific terminology coinage (`Dīwān al-Iṣṭilāḥ`) & governed scientific machine translation (`TG → ATC → ADT`, `F ⊣ U` discourse folding) (`Apache-2.0` code / `CC BY 4.0` `diwan_public.db`).
- `maqayis/misbar` (`v1.0.0`): Deterministic workflow trace spine, zero-dependency capture, and blind RTL/LTR human calibration & review queues (`Apache-2.0`).
- `maqayis/manhaj` (`v1.0.0`): Interactive bilingual computational linguistics & mathematical foundations academy (`Apache-2.0` runtime / `CC BY 4.0` curriculum).

## Zenodo Archival Deposits (`CC BY 4.0`)

1. **Handasat al-Lisān al-ʿArabī (v2.0)**: Mathematical & Computational Foundations of Arabic (`publication`, `v2.0.0`).
2. **Tarkīb & Jadwal Zero-LLM Arabic Lexical-Geometric Atlas (v2.0)**: OpenITI-derived clean-room root, subroot, *muhmal*, and `Iqtirān` datasets (`dataset`, `v2.0.0`).
3. **Dīwān al-Iṣṭilāḥ (v1.0)**: Governed Arabic Scientific Terminology Ledger & Bilingual Paper Corpus (`dataset`, `v1.0.0`).

## Machine-Readable Static JSON Shards (`GET`, `CC BY 4.0`)

- `/static-api/book/toc.json`, `/static-api/book/glossary.json`, `/static-api/book/experiments.json`, `/static-api/book/monographs.json`
- `/static-api/jadwal/overview.json`, `/static-api/jadwal/subroots.json`, `/static-api/jadwal/furuq.json`, `/static-api/jadwal/muhmal.json`, `/static-api/jadwal/shards/b00.json`..`b27.json`
- `/static-api/iqtiran/atoms.json`
- `/static-api/diwan/stats.json`, `/static-api/diwan/terms.json`
- `/static-api/awn4/overview.json`, `/static-api/awn4/lex_index/ar_00.json`..`ar_27.json`, `/static-api/awn4/lex_index/en_a.json`..`en_z.json`, `/static-api/awn4/shards/s00.json`..`s7f.json`
- `/static-api/tayf/arena_bundle.json`
- `/static-api/papers/index.json`, `/static-api/papers/colbert-2020.json`, `/static-api/papers/lora-2021.json`, `/static-api/papers/multilingual-e5-2024.json`

## Edge Observatory Public API (`/api/v1/*`)

- `GET /api/v1/health` — Service health, zero-LLM-custody guarantee, consent grantee (`Salah Abdo and successor entity`), and public counters.
- `GET /api/v1/diwan/proposals` & `POST /api/v1/diwan/propose` — Community scientific term proposals validated by deterministic `Σ₂₈` OCP phonotactic gate.
- `GET /api/v1/benchmarks/leaderboard` & `POST /api/v1/benchmarks/submit` — Open benchmark baselines & reproducible community receipts.
- `POST /api/v1/traces` & `DELETE /api/v1/traces/:anon_id` — Opt-in double-scrubbed research telemetry with 1-click HMAC right-to-erasure.