Arabic Dialect Ontology — Prototype

A SwiftData-based ontology for Arabic dialects supporting localization and translation work.

Quick Start

1. Create Xcode Project

Requirements: macOS 14 (Sonoma) or later, Xcode 15+, Swift 5.9+

1. File → New → Project
2. Select macOS → App
3. Product Name: ArabicOntology
4. Interface: SwiftUI
5. Language: Swift
6. Storage: None (we configure SwiftData manually)

2. Add Source Files

Drag the three folders into Xcode's Project Navigator:

1. Drag Models/, Importer/, App/ into the project
2. In the dialog: select Create groups
3. Ensure Add to targets has ArabicOntology checked

All 11 Swift files should now appear in your project with target membership set to the app.

3. Prepare Data Folder

Collect CSV files in one folder:

data/
├── Qabas-dataset.csv             # Required — lemmas and roots
├── Concepts.csv                  # Required — concept definitions
├── Relations.csv                 # Required — concept hierarchy
├── Baladi-dataset.csv            # Optional — Lebanese forms
├── Baladi_RowText_sentences.csv  # Optional — Lebanese sentences
├── Nabra-dataset.csv             # Optional — Syrian forms
├── Nabra_RowText_sentences.csv   # Optional — Syrian sentences
├── Curras-dataset.csv            # Optional — Palestinian forms
├── Curras_RowText_sentences.csv  # Optional — Palestinian sentences
├── Lisan-Iraqi-dataset.csv       # Optional — Iraqi forms
├── Lisan-Iraqi_RowText_sentences.csv
├── Lisan-Libyan-dataset.csv      # Optional — Libyan forms
├── Lisan-Libyan_RowText_sentences.csv
├── Lisan-Sudanese-dataset.csv    # Optional — Sudanese forms
├── Lisan-Sudanese_RowText_sentences.csv
├── Lisan-Yemeni-dataset.csv      # Optional — Yemeni forms
└── Lisan-Yemeni_RowText_sentences.csv

Note: The three required files give you the lexicon (58K lemmas, 16K roots, 14K concepts). Missing corpus files are skipped with warnings — but without them you won't have Forms, Sentences, or dialect correspondences. For comparing translations, include all corpus files.

4. Run App and Import

1. Build and run
2. Click "Import Data"
3. Select your data folder
4. Wait for import (~10-15 min for full dataset)
5. Search and explore

Database location: ~/Library/Application Support/ArabicOntology/ArabicOntology.store

To reset and re-import, quit the app and delete this file.

Project Structure

ArabicOntology/
├── Models/                 # SwiftData entities
│   ├── Concept.swift
│   ├── Root.swift
│   ├── Dialect.swift
│   ├── Lemma.swift
│   ├── Form.swift
│   ├── Sentence.swift
│   └── ArabicNormalizer.swift
├── Importer/
│   ├── ImportService.swift # CSV → SwiftData
│   └── CSVParser.swift
└── App/
    ├── ArabicOntologyApp.swift
    ├── ContentView.swift
    └── QueryService.swift

Entity Summary

Entity	Count	Description
Concept	~14,000	Meaning units with Arabic/English synsets
Root	~16,000	Consonantal skeletons (ك ت ب)
Lemma	~58,000	Dictionary headwords
Form	~1,270,000	Attested tokens in corpora
Sentence	~56,000	Usage contexts
Dialect	8	MSA + 7 regional varieties

Design Principles

1. MSA is a dialect — treated as one variety among equals
2. Concept and Root are independent dimensions — query through either
3. Correspondence is symmetric — no dialect is privileged
4. Open-world polysemy — one lemma can express multiple concepts
5. Register ≠ Dialect — Qabas language field indicates register (MSA/colloquial/foreign), not dialect; all map to MSA for queries; actual dialect comes from corpus provenance

Citations

See the Design Proposal document for full citation requirements for:

• Qabas lexicon (Birzeit University)
• Arabic Ontology (Birzeit University)
• Dialect corpora (Birzeit SinaLab)