README.md

---
language: ug
language_name: Uyghur
language_family: turkic_other
tags:
  - wikilangs
  - nlp
  - tokenizer
  - embeddings
  - n-gram
  - markov
  - wikipedia
  - feature-extraction
  - sentence-similarity
  - tokenization
  - n-grams
  - markov-chain
  - text-mining
  - fasttext
  - babelvec
  - vocabulous
  - vocabulary
  - monolingual
  - family-turkic_other
license: mit
library_name: wikilangs
pipeline_tag: text-generation
datasets:
  - omarkamali/wikipedia-monthly
dataset_info:
  name: wikipedia-monthly
  description: Monthly snapshots of Wikipedia articles across 300+ languages
metrics:
  - name: best_compression_ratio
    type: compression
    value: 4.777
  - name: best_isotropy
    type: isotropy
    value: 0.8332
  - name: vocabulary_size
    type: vocab
    value: 0
generated: 2026-01-11
---

# Uyghur - Wikilangs Models
## Comprehensive Research Report & Full Ablation Study

This repository contains NLP models trained and evaluated by Wikilangs, specifically on **Uyghur** Wikipedia data.
We analyze tokenizers, n-gram models, Markov chains, vocabulary statistics, and word embeddings.

## 📋 Repository Contents

### Models & Assets

- Tokenizers (8k, 16k, 32k, 64k)
- N-gram models (2, 3, 4, 5-gram)
- Markov chains (context of 1, 2, 3, 4 and 5)
- Subword N-gram and Markov chains
- Embeddings in various sizes and dimensions (aligned and unaligned)
- Language Vocabulary
- Language Statistics

![Performance Dashboard](visualizations/performance_dashboard.png)

### Analysis and Evaluation

- [1. Tokenizer Evaluation](#1-tokenizer-evaluation)
- [2. N-gram Model Evaluation](#2-n-gram-model-evaluation)
- [3. Markov Chain Evaluation](#3-markov-chain-evaluation)
- [4. Vocabulary Analysis](#4-vocabulary-analysis)
- [5. Word Embeddings Evaluation](#5-word-embeddings-evaluation)
- [6. Morphological Analysis (Experimental)](#6--morphological-analysis-experimental)
- [7. Summary & Recommendations](#7-summary--recommendations)
- [Metrics Glossary](#appendix-metrics-glossary--interpretation-guide)
- [Visualizations Index](#visualizations-index)

---
## 1. Tokenizer Evaluation

![Tokenizer Compression](visualizations/tokenizer_compression.png)

![Tokenizer Fertility](visualizations/tokenizer_fertility.png)

![Tokenizer OOV](visualizations/tokenizer_oov.png)

![Total Tokens](visualizations/tokenizer_total_tokens.png)

### Results

| Vocab Size | Compression | Avg Token Len | UNK Rate | Total Tokens |
|------------|-------------|---------------|----------|--------------|
| **8k** | 3.736x | 3.74 | 0.1104% | 494,604 |
| **16k** | 4.184x | 4.19 | 0.1236% | 441,635 |
| **32k** | 4.530x | 4.54 | 0.1339% | 407,907 |
| **64k** | 4.777x 🏆 | 4.78 | 0.1412% | 386,795 |

### Tokenization Examples

Below are sample sentences tokenized with each vocabulary size:

**Sample 1:** `تۇنىسدىكى ئۇيغۇرلارنىڭ سانى 10 ئەتراپىدا بولۇپ، كۆزگە كۆرۈنگەن شەخىسلەر : مەنبەل...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁تۇ نىس دىكى ▁ئۇيغۇرلارنىڭ ▁سانى ▁ 1 0 ▁ئەتراپىدا ▁بولۇپ ... (+6 more)` | 16 |
| 16k | `▁تۇنىس دىكى ▁ئۇيغۇرلارنىڭ ▁سانى ▁ 1 0 ▁ئەتراپىدا ▁بولۇپ ، ... (+5 more)` | 15 |
| 32k | `▁تۇنىس دىكى ▁ئۇيغۇرلارنىڭ ▁سانى ▁ 1 0 ▁ئەتراپىدا ▁بولۇپ ، ... (+5 more)` | 15 |
| 64k | `▁تۇنىس دىكى ▁ئۇيغۇرلارنىڭ ▁سانى ▁ 1 0 ▁ئەتراپىدا ▁بولۇپ ، ... (+5 more)` | 15 |

**Sample 2:** `مۈشۈك ئائىلىسى بەلگە جايلاشماق ئادەت ئاۋۇماق قونالغۇ ئوزۇقلىنىش خۇسۇسىيىتى كەنجى...`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁مۈشۈك ▁ئائىلىسى ▁بەلگە ▁جايلاشماق ▁ئادەت ▁ئاۋۇماق ▁قونالغۇ ▁ئوزۇقلىنىش ▁خۇسۇسىيىتى ▁كەنجى ... (+5 more)` | 15 |
| 16k | `▁مۈشۈك ▁ئائىلىسى ▁بەلگە ▁جايلاشماق ▁ئادەت ▁ئاۋۇماق ▁قونالغۇ ▁ئوزۇقلىنىش ▁خۇسۇسىيىتى ▁كەنجى ... (+5 more)` | 15 |
| 32k | `▁مۈشۈك ▁ئائىلىسى ▁بەلگە ▁جايلاشماق ▁ئادەت ▁ئاۋۇماق ▁قونالغۇ ▁ئوزۇقلىنىش ▁خۇسۇسىيىتى ▁كەنجى ... (+5 more)` | 15 |
| 64k | `▁مۈشۈك ▁ئائىلىسى ▁بەلگە ▁جايلاشماق ▁ئادەت ▁ئاۋۇماق ▁قونالغۇ ▁ئوزۇقلىنىش ▁خۇسۇسىيىتى ▁كەنجى ... (+5 more)` | 15 |

**Sample 3:** `ئاپتونوم رايونلۇق تۇرالغۇ ۋە شەھەر - يېزا قۇرۇلۇشى نازارىتى قۇرۇلدى. مەنبەلەر`

| Vocab | Tokens | Count |
|-------|--------|-------|
| 8k | `▁ئاپتونوم ▁رايونلۇق ▁تۇر الغۇ ▁ۋە ▁شەھەر ▁- ▁يېزا ▁قۇرۇلۇشى ▁ناز ... (+4 more)` | 14 |
| 16k | `▁ئاپتونوم ▁رايونلۇق ▁تۇرالغۇ ▁ۋە ▁شەھەر ▁- ▁يېزا ▁قۇرۇلۇشى ▁نازارىتى ▁قۇرۇلدى ... (+2 more)` | 12 |
| 32k | `▁ئاپتونوم ▁رايونلۇق ▁تۇرالغۇ ▁ۋە ▁شەھەر ▁- ▁يېزا ▁قۇرۇلۇشى ▁نازارىتى ▁قۇرۇلدى ... (+2 more)` | 12 |
| 64k | `▁ئاپتونوم ▁رايونلۇق ▁تۇرالغۇ ▁ۋە ▁شەھەر ▁- ▁يېزا ▁قۇرۇلۇشى ▁نازارىتى ▁قۇرۇلدى ... (+2 more)` | 12 |


### Key Findings

- **Best Compression:** 64k achieves 4.777x compression
- **Lowest UNK Rate:** 8k with 0.1104% unknown tokens
- **Trade-off:** Larger vocabularies improve compression but increase model size
- **Recommendation:** 32k vocabulary provides optimal balance for production use

---
## 2. N-gram Model Evaluation

![N-gram Perplexity](visualizations/ngram_perplexity.png)

![N-gram Unique](visualizations/ngram_unique.png)

![N-gram Coverage](visualizations/ngram_coverage.png)

### Results

| N-gram | Variant | Perplexity | Entropy | Unique N-grams | Top-100 Coverage | Top-1000 Coverage |
|--------|---------|------------|---------|----------------|------------------|-------------------|
| **2-gram** | Word | 31,118 | 14.93 | 65,522 | 7.0% | 22.8% |
| **2-gram** | Subword | 453 🏆 | 8.82 | 10,807 | 59.0% | 94.8% |
| **3-gram** | Word | 26,893 | 14.71 | 61,707 | 9.8% | 27.8% |
| **3-gram** | Subword | 3,539 | 11.79 | 73,196 | 23.7% | 64.2% |
| **4-gram** | Word | 113,663 | 16.79 | 204,980 | 5.9% | 16.1% |
| **4-gram** | Subword | 17,568 | 14.10 | 306,560 | 10.8% | 35.0% |
| **5-gram** | Word | 102,988 | 16.65 | 180,010 | 6.1% | 16.5% |
| **5-gram** | Subword | 57,887 | 15.82 | 660,169 | 6.0% | 21.1% |

### Top 5 N-grams by Size

**2-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `font family` | 1,559 |
| 2 | `style direction` | 1,545 |
| 3 | `div style` | 1,544 |
| 4 | `قاماق جازاسى` | 1,444 |
| 5 | `يەنە بىر` | 1,422 |

**3-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `div style direction` | 1,538 |
| 2 | `مۇددەتلىك قاماق جازاسى` | 1,253 |
| 3 | `style direction rtl` | 1,242 |
| 4 | `direction rtl font` | 1,239 |
| 5 | `rtl font family` | 1,239 |

**4-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `style direction rtl font` | 1,239 |
| 2 | `direction rtl font family` | 1,239 |
| 3 | `div style direction rtl` | 1,237 |
| 4 | `تۆۋەن مۇددەتلىك قاماق جازاسى` | 971 |
| 5 | `يىلدىن تۆۋەن مۇددەتلىك قاماق` | 963 |

**5-grams (Word):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `style direction rtl font family` | 1,239 |
| 2 | `div style direction rtl font` | 1,234 |
| 3 | `يىلدىن تۆۋەن مۇددەتلىك قاماق جازاسى` | 955 |
| 4 | `tom microsoft uighur uyghur ekran` | 802 |
| 5 | `ukij tuz tom microsoft uighur` | 802 |

**2-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ ئ` | 601,928 |
| 2 | `ى _` | 466,826 |
| 3 | `ل ى` | 449,775 |
| 4 | `ن ى` | 375,089 |
| 5 | `ى ل` | 331,034 |

**3-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `_ ئ ا` | 151,323 |
| 2 | `ن ى ڭ` | 143,621 |
| 3 | `ى ڭ _` | 142,034 |
| 4 | `ى ن ى` | 138,625 |
| 5 | `ى ل ى` | 138,069 |

**4-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ن ى ڭ _` | 135,065 |
| 2 | `_ ب و ل` | 68,419 |
| 3 | `غ ا ن _` | 58,174 |
| 4 | `د ى ن _` | 57,430 |
| 5 | `ل ى ر ى` | 56,859 |

**5-grams (Subword):**

| Rank | N-gram | Count |
|------|--------|-------|
| 1 | `ى ن ى ڭ _` | 44,317 |
| 2 | `_ ق ى ل ى` | 35,323 |
| 3 | `ن ى ڭ _ ئ` | 33,703 |
| 4 | `د ى ك ى _` | 29,476 |
| 5 | `_ ب ى ر _` | 29,261 |


### Key Findings

- **Best Perplexity:** 2-gram (subword) with 453
- **Entropy Trend:** Decreases with larger n-grams (more predictable)
- **Coverage:** Top-1000 patterns cover ~21% of corpus
- **Recommendation:** 4-gram or 5-gram for best predictive performance

---
## 3. Markov Chain Evaluation

![Markov Entropy](visualizations/markov_entropy.png)

![Markov Contexts](visualizations/markov_contexts.png)

![Markov Branching](visualizations/markov_branching.png)

### Results

| Context | Variant | Avg Entropy | Perplexity | Branching Factor | Unique Contexts | Predictability |
|---------|---------|-------------|------------|------------------|-----------------|----------------|
| **1** | Word | 0.8072 | 1.750 | 5.72 | 320,664 | 19.3% |
| **1** | Subword | 1.1829 | 2.270 | 8.19 | 4,322 | 0.0% |
| **2** | Word | 0.2037 | 1.152 | 1.42 | 1,831,690 | 79.6% |
| **2** | Subword | 0.7055 | 1.631 | 4.46 | 35,371 | 29.5% |
| **3** | Word | 0.0501 | 1.035 | 1.07 | 2,595,280 | 95.0% |
| **3** | Subword | 0.7110 | 1.637 | 3.49 | 157,874 | 28.9% |
| **4** | Word | 0.0173 🏆 | 1.012 | 1.02 | 2,783,101 | 98.3% |
| **4** | Subword | 0.5528 | 1.467 | 2.45 | 550,624 | 44.7% |

### Generated Text Samples (Word-based)

Below are text samples generated from each word-based Markov chain model:

**Context Size 1:**

1. `ۋە شەكلىمۇ ئۆزگىچە ھەر دەرىجىلىك قوغدىلىدىغان مەدەنىيەت گۇرۇپپىلىرى ديكى ۋە ئىزچىللىقىدۇر يەنگىن 85 ...`
2. `بىر خىزمەت ئورنىڭىزدىن تۇرالامسىز ــ ئۇيغۇر قاغانلىقىغا ئەلچى ئەۋەتتى بۇ كىشى ئىسمى بولغان جورنالىست...`
3. `بىلەن ئۇيغۇر ئاپتونوم جايلارنىڭ ئاپتونومىيە ئورگانلىرى تەسىس قىلىنغان ئۇ يەردىكى ۋەتەنسىز ئادەم توپل...`

**Context Size 2:**

1. `font family ukij tuz tom microsoft uighur uyghur ekran arial unicode ms alpida_unicode system ukij t...`
2. `style direction rtl font family tahoma ئۇمۇمىي ئىسمى سىمۋولى نومۇرى ئىلمېنىت كاتېگورىيىسى گۇرۇپپىسى ...`
3. `div style direction rtl font family ukij tuz tom microsoft uighur uyghur ekran arial unicode ms row`

**Context Size 3:**

1. `div style direction rtl font family tahoma wirginiye shitati bolsa amérika qoshma shtatliri dikibixa...`
2. `مۇددەتلىك قاماق جازاسى ياكى تۇتۇپ تۇرۇپ ئەمگەككە سېلىش جازاسى بېرىلىدۇ ئاقىۋىتى پەۋقۇلئاددە ئېغىر بو...`
3. `style direction rtl font family tahoma misori shitati bolsa amérika qoshma shtatliri dikibixahar nop...`

**Context Size 4:**

1. `style direction rtl font family alkatip tor alpida_unicode system ukij tuz tom microsoft uighur uygh...`
2. `direction rtl font family alkatip tor alpida_unicode system ukij tuz tom microsoft uighur uyghur ekr...`
3. `div style direction rtl font family ukij tuz basma microsoft uighur uyghur ekran arial unicode ms بې...`


### Generated Text Samples (Subword-based)

Below are text samples generated from each subword-based Markov chain model:

**Context Size 1:**

1. `_سەي_پان-_«ﺗﻪﻧﻰ_`
2. `ى،_شلىسارىرىقاھر`
3. `ابىشكىلىرلىقىئۇ_`

**Context Size 2:**

1. `_ئىق_كېلىقىليەت_ي`
2. `ى_ۋە_ئەككىن_يولغا`
3. `لىك_ئادىغا_يا_مېر`

**Context Size 3:**

1. `_ئاياغا_ۋە_ھازى_يو`
2. `نىڭ_يۈرۈپ_كەلتۈركچ`
3. `ىڭ_تۆۋەتەنتتە_يۈرى`

**Context Size 4:**

1. `نىڭ_نەتىجىلىك_ماۋۇ_`
2. `_بولسا_نىسبەت_قىلىد`
3. `غان_ھايۋانات_دولقۇن`


### Key Findings

- **Best Predictability:** Context-4 (word) with 98.3% predictability
- **Branching Factor:** Decreases with context size (more deterministic)
- **Memory Trade-off:** Larger contexts require more storage (550,624 contexts)
- **Recommendation:** Context-3 or Context-4 for text generation

---
## 4. Vocabulary Analysis

![Zipf's Law](visualizations/zipf_law.png)

![Top Words](visualizations/top20_words.png)

![Coverage Curve](visualizations/vocab_coverage.png)

### Statistics

| Metric | Value |
|--------|-------|
| Vocabulary Size | 135,750 |
| Total Tokens | 3,012,380 |
| Mean Frequency | 22.19 |
| Median Frequency | 3 |
| Frequency Std Dev | 244.00 |

### Most Common Words

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | ۋە | 48,299 |
| 2 | بىر | 30,257 |
| 3 | بىلەن | 28,269 |
| 4 | بۇ | 23,896 |
| 5 | بولۇپ | 15,446 |
| 6 | بولغان | 12,330 |
| 7 | ياكى | 11,679 |
| 8 | ئۇيغۇر | 11,561 |
| 9 | ئۇ | 10,650 |
| 10 | دەپ | 9,455 |

### Least Common Words (from vocabulary)

| Rank | Word | Frequency |
|------|------|-----------|
| 1 | hâmid | 2 |
| 2 | arî | 2 |
| 3 | alâ | 2 |
| 4 | tevhîd | 2 |
| 5 | kitâbu | 2 |
| 6 | âsım | 2 |
| 7 | بىرقانداق | 2 |
| 8 | گەنبىيەن | 2 |
| 9 | توغرامچىسى | 2 |
| 10 | باۋزا | 2 |

### Zipf's Law Analysis

| Metric | Value |
|--------|-------|
| Zipf Coefficient | 0.9498 |
| R² (Goodness of Fit) | 0.987707 |
| Adherence Quality | **excellent** |

### Coverage Analysis

| Top N Words | Coverage |
|-------------|----------|
| Top 100 | 18.6% |
| Top 1,000 | 43.9% |
| Top 5,000 | 66.4% |
| Top 10,000 | 75.4% |

### Key Findings

- **Zipf Compliance:** R²=0.9877 indicates excellent adherence to Zipf's law
- **High Frequency Dominance:** Top 100 words cover 18.6% of corpus
- **Long Tail:** 125,750 words needed for remaining 24.6% coverage

---
## 5. Word Embeddings Evaluation

![Embedding Isotropy](visualizations/embedding_isotropy.png)

![Similarity Matrix](visualizations/embedding_similarity.png)

![t-SNE Words](visualizations/tsne_words.png)

![t-SNE Sentences](visualizations/tsne_sentences.png)


### 5.1 Cross-Lingual Alignment

![Alignment Quality](visualizations/embedding_alignment_quality.png)

![Multilingual t-SNE](visualizations/embedding_tsne_multilingual.png)


### 5.2 Model Comparison

| Model | Dimension | Isotropy | Semantic Density | Alignment R@1 | Alignment R@10 |
|-------|-----------|----------|------------------|---------------|----------------|
| **mono_32d** | 32 | 0.8332 🏆 | 0.3495 | N/A | N/A |
| **mono_64d** | 64 | 0.8220 | 0.2582 | N/A | N/A |
| **mono_128d** | 128 | 0.8294 | 0.1749 | N/A | N/A |
| **aligned_32d** | 32 | 0.8332 | 0.3542 | 0.0200 | 0.1300 |
| **aligned_64d** | 64 | 0.8220 | 0.2639 | 0.0460 | 0.2060 |
| **aligned_128d** | 128 | 0.8294 | 0.1693 | 0.0880 | 0.2980 |

### Key Findings

- **Best Isotropy:** mono_32d with 0.8332 (more uniform distribution)
- **Semantic Density:** Average pairwise similarity of 0.2617. Lower values indicate better semantic separation.
- **Alignment Quality:** Aligned models achieve up to 8.8% R@1 in cross-lingual retrieval.
- **Recommendation:** 128d aligned for best cross-lingual performance

---
## 6.  Morphological Analysis (Experimental)

This section presents an automated morphological analysis derived from the statistical divergence between word-level and subword-level models. By analyzing where subword predictability spikes and where word-level coverage fails, we can infer linguistic structures without supervised data.

### 6.1 Productivity & Complexity

| Metric | Value | Interpretation | Recommendation |
|--------|-------|----------------|----------------|
| Productivity Index | **5.000** | High morphological productivity | Reliable analysis |
| Idiomaticity Gap | **-0.217** | Low formulaic content | - |

### 6.2 Affix Inventory (Productive Units)

These are the most productive prefixes and suffixes identified by sampling the vocabulary for global substitutability patterns. A unit is considered an affix if stripping it leaves a valid stem that appears in other contexts.

#### Productive Prefixes
| Prefix | Examples |
|--------|----------|
| `-ئ` | ئۇقۇتۇش, ئۇبباد, ئاپتوبۇسنىڭ |
| `-ئا` | ئاپتوبۇسنىڭ, ئاياغنى, ئاقساشقا |
| `-ت` | تەلەيسىزلىكىدىن, تاتلىقنىڭ, توشمىغانلار |
| `-م` | مۇشتەرى, مەنبئەدىن, مۇزلار |
| `-ئە` | ئەتتۈرىدۇكى, ئەﻫﻤﯩﻴﻪﺗﻠﯩﻚ, ئەﮬلى |
| `-ك` | كېچىلىرى, كۋانتى, كارتىنىڭ |
| `-ب` | بەرداشلىق, بەئەينى, بېكىتمىگەن |
| `-س` | سەرەتان, سەيياھنىڭ, سوغۇقچان |

#### Productive Suffixes
| Suffix | Examples |
|--------|----------|
| `-ى` | كېچىلىرى, كۋانتى, ھېسىياتلىرى |
| `-ىڭ` | شەكلىنىڭ, سەيياھنىڭ, ئاپتوبۇسنىڭ |
| `-ن` | چاپلانغان, سەرەتان, غەزەپتىن |
| `-نى` | شىركەتلەرنى, نوچىنى, بەئەينى |
| `-ڭ` | شەكلىنىڭ, سەيياھنىڭ, ئاپتوبۇسنىڭ |
| `-ىن` | غەزەپتىن, دەرەختىن, تەلەيسىزلىكىدىن |
| `-ا` | لايكا, ئىستىقبالغا, ھېساپلىغاندا |
| `-ە` | ئۆزلەشتۈرۈشىگە, سەۋىيىگىچە, كەچلەردە |

### 6.3 Bound Stems (Lexical Roots)

Bound stems are high-frequency subword units that are semantically cohesive but rarely appear as standalone words. These often correspond to the 'core' of a word that requires inflection or derivation to be valid.

| Stem | Cohesion | Substitutability | Examples |
|------|----------|------------------|----------|
| `رىنى` | 2.13x | 126 contexts | برىنى, يىرىنى, نارىنى |
| `ىدىك` | 2.04x | 131 contexts | خىدىكى, سىدىكى, ئىدىكى |
| `لارن` | 1.92x | 162 contexts | لارنى, لارنىڭ, سۇلارنى |
| `ىرىن` | 2.05x | 80 contexts | سىرىن, شىرىن, بىرىن |
| `ىلىش` | 1.62x | 247 contexts | بىلىش, ئىلىش, تىلىش |
| `ارنى` | 1.96x | 72 contexts | لارنى, قارنى, بارنى |
| `يغۇر` | 2.39x | 29 contexts | ئۇيغۇر, ئويغۇر, ئ‍ۇيغۇر |
| `انلى` | 1.60x | 116 contexts | شانلى, جانلىق, دانلىد |
| `دىغا` | 2.14x | 34 contexts | دىغار, مودىغا, سودىغا |
| `ىلىق` | 1.54x | 132 contexts | سىلىق, بىلىق, يىلىق |
| `رنىڭ` | 2.14x | 29 contexts | ئەرنىڭ, كورنىڭ, يەرنىڭ |
| `ىغان` | 1.59x | 94 contexts | پىغان, جىغان, قلىغان |

### 6.4 Affix Compatibility (Co-occurrence)

This table shows which prefixes and suffixes most frequently co-occur on the same stems, revealing the 'stacking' rules of the language's morphology.

| Prefix | Suffix | Frequency | Examples |
|--------|--------|-----------|----------|
| `-ئ` | `-ى` | 327 words | ئۇزىرىشى, ئۇلاغلىرىنى |
| `-ت` | `-ى` | 181 words | تاۋارلارنى, تاۋابىئاتلىرى |
| `-ئ` | `-ن` | 176 words | ئەرەبچىدىن, ئىپادىلەيدىغان |
| `-ئ` | `-ا` | 164 words | ئاقسارايغا, ئۇلاپلا |
| `-ئ` | `-ڭ` | 154 words | ئۇنىۋېرسىتېتنىڭ, ئىلتىماسنىڭ |
| `-ئ` | `-ىڭ` | 151 words | ئۇنىۋېرسىتېتنىڭ, ئىلتىماسنىڭ |
| `-ئ` | `-نى` | 136 words | ئۇلاغلىرىنى, ئەترەتنى |
| `-ئ` | `-ە` | 117 words | ئىبرانىيچە, ئۆستۈرۈشتە |
| `-م` | `-ى` | 113 words | مايورى, مەسۇلاتلىرىنى |
| `-ك` | `-ى` | 100 words | كىنوچىلىقى, كۈنپېتىشتىكى |

### 6.5 Recursive Morpheme Segmentation

Using **Recursive Hierarchical Substitutability**, we decompose complex words into their constituent morphemes. This approach handles nested affixes (e.g., `prefix-prefix-root-suffix`).

| Word | Suggested Split | Confidence | Stem |
|------|-----------------|------------|------|
| پەرقلەنسىمۇ | **`پەرقلەن-سى-مۇ`** | 7.5 | `سى` |
| لەپتىننىڭ | **`لەپتىن-ن-ىڭ`** | 7.5 | `ن` |
| تەشكىلاتنىڭ | **`تەشكىلات-ن-ىڭ`** | 7.5 | `ن` |
| تاشيولنىڭ | **`تاشيول-ن-ىڭ`** | 7.5 | `ن` |
| ئىستاكاننىڭ | **`ئىستاكان-ن-ىڭ`** | 7.5 | `ن` |
| ئۆستۈرۈشنىڭ | **`ئۆستۈرۈش-ن-ىڭ`** | 7.5 | `ن` |
| ھۈجەيرىسىنى | **`ھۈجەيرى-سى-نى`** | 7.5 | `سى` |
| دوستويېۋىسكىي | **`دوستويېۋىس-كى-ي`** | 7.5 | `كى` |
| قۇتۇلۇشنىڭ | **`قۇتۇلۇش-ن-ىڭ`** | 7.5 | `ن` |
| ئىقتىدارىمنىڭ | **`ئىقتىدارىم-ن-ىڭ`** | 7.5 | `ن` |
| پۇقرالىقىغا | **`پۇقرالىق-ى-غا`** | 6.0 | `پۇقرالىق` |
| ئوخشىماسلىقىغا | **`ئوخشىماسلىق-ى-غا`** | 6.0 | `ئوخشىماسلىق` |
| ئاسساسلىق | **`ئا-س-ساسلىق`** | 6.0 | `ساسلىق` |
| ئادەمىنىڭ | **`ئادەم-ىن-ىڭ`** | 6.0 | `ئادەم` |
| ئۈچۈنچىسى | **`ئۈچۈن-چى-سى`** | 6.0 | `ئۈچۈن` |

### 6.6 Linguistic Interpretation

> **Automated Insight:**
The language Uyghur shows high morphological productivity. The subword models are significantly more efficient than word models, suggesting a rich system of affixation or compounding.

---
## 7. Summary & Recommendations

![Performance Dashboard](visualizations/performance_dashboard.png)

### Production Recommendations

| Component | Recommended | Rationale |
|-----------|-------------|-----------|
| Tokenizer | **64k BPE** | Best compression (4.78x) |
| N-gram | **2-gram** | Lowest perplexity (453) |
| Markov | **Context-4** | Highest predictability (98.3%) |
| Embeddings | **100d** | Balanced semantic capture and isotropy |


---
## Appendix: Metrics Glossary & Interpretation Guide

This section provides definitions, intuitions, and guidance for interpreting the metrics used throughout this report.

### Tokenizer Metrics

**Compression Ratio**
> *Definition:* The ratio of characters to tokens (chars/token). Measures how efficiently the tokenizer represents text.
>
> *Intuition:* Higher compression means fewer tokens needed to represent the same text, reducing sequence lengths for downstream models. A 3x compression means ~3 characters per token on average.
>
> *What to seek:* Higher is generally better for efficiency, but extremely high compression may indicate overly aggressive merging that loses morphological information.

**Average Token Length (Fertility)**
> *Definition:* Mean number of characters per token produced by the tokenizer.
>
> *Intuition:* Reflects the granularity of tokenization. Longer tokens capture more context but may struggle with rare words; shorter tokens are more flexible but increase sequence length.
>
> *What to seek:* Balance between 2-5 characters for most languages. Arabic/morphologically-rich languages may benefit from slightly longer tokens.

**Unknown Token Rate (OOV Rate)**
> *Definition:* Percentage of tokens that map to the unknown/UNK token, indicating words the tokenizer cannot represent.
>
> *Intuition:* Lower OOV means better vocabulary coverage. High OOV indicates the tokenizer encounters many unseen character sequences.
>
> *What to seek:* Below 1% is excellent; below 5% is acceptable. BPE tokenizers typically achieve very low OOV due to subword fallback.

### N-gram Model Metrics

**Perplexity**
> *Definition:* Measures how "surprised" the model is by test data. Mathematically: 2^(cross-entropy). Lower values indicate better prediction.
>
> *Intuition:* If perplexity is 100, the model is as uncertain as if choosing uniformly among 100 options at each step. A perplexity of 10 means effectively choosing among 10 equally likely options.
>
> *What to seek:* Lower is better. Perplexity decreases with larger n-grams (more context). Values vary widely by language and corpus size.

**Entropy**
> *Definition:* Average information content (in bits) needed to encode the next token given the context. Related to perplexity: perplexity = 2^entropy.
>
> *Intuition:* High entropy means high uncertainty/randomness; low entropy means predictable patterns. Natural language typically has entropy between 1-4 bits per character.
>
> *What to seek:* Lower entropy indicates more predictable text patterns. Entropy should decrease as n-gram size increases.

**Coverage (Top-K)**
> *Definition:* Percentage of corpus occurrences explained by the top K most frequent n-grams.
>
> *Intuition:* High coverage with few patterns indicates repetitive/formulaic text; low coverage suggests diverse vocabulary usage.
>
> *What to seek:* Depends on use case. For language modeling, moderate coverage (40-60% with top-1000) is typical for natural text.

### Markov Chain Metrics

**Average Entropy**
> *Definition:* Mean entropy across all contexts, measuring average uncertainty in next-word prediction.
>
> *Intuition:* Lower entropy means the model is more confident about what comes next. Context-1 has high entropy (many possible next words); Context-4 has low entropy (few likely continuations).
>
> *What to seek:* Decreasing entropy with larger context sizes. Very low entropy (<0.1) indicates highly deterministic transitions.

**Branching Factor**
> *Definition:* Average number of unique next tokens observed for each context.
>
> *Intuition:* High branching = many possible continuations (flexible but uncertain); low branching = few options (predictable but potentially repetitive).
>
> *What to seek:* Branching factor should decrease with context size. Values near 1.0 indicate nearly deterministic chains.

**Predictability**
> *Definition:* Derived metric: (1 - normalized_entropy) × 100%. Indicates how deterministic the model's predictions are.
>
> *Intuition:* 100% predictability means the next word is always certain; 0% means completely random. Real text falls between these extremes.
>
> *What to seek:* Higher predictability for text generation quality, but too high (>98%) may produce repetitive output.

### Vocabulary & Zipf's Law Metrics

**Zipf's Coefficient**
> *Definition:* The slope of the log-log plot of word frequency vs. rank. Zipf's law predicts this should be approximately -1.
>
> *Intuition:* A coefficient near -1 indicates the corpus follows natural language patterns where a few words are very common and most words are rare.
>
> *What to seek:* Values between -0.8 and -1.2 indicate healthy natural language distribution. Deviations may suggest domain-specific or artificial text.

**R² (Coefficient of Determination)**
> *Definition:* Measures how well the linear fit explains the frequency-rank relationship. Ranges from 0 to 1.
>
> *Intuition:* R² near 1.0 means the data closely follows Zipf's law; lower values indicate deviation from expected word frequency patterns.
>
> *What to seek:* R² > 0.95 is excellent; > 0.99 indicates near-perfect Zipf adherence typical of large natural corpora.

**Vocabulary Coverage**
> *Definition:* Cumulative percentage of corpus tokens accounted for by the top N words.
>
> *Intuition:* Shows how concentrated word usage is. If top-100 words cover 50% of text, the corpus relies heavily on common words.
>
> *What to seek:* Top-100 covering 30-50% is typical. Higher coverage indicates more repetitive text; lower suggests richer vocabulary.

### Word Embedding Metrics

**Isotropy**
> *Definition:* Measures how uniformly distributed vectors are in the embedding space. Computed as the ratio of minimum to maximum singular values.
>
> *Intuition:* High isotropy (near 1.0) means vectors spread evenly in all directions; low isotropy means vectors cluster in certain directions, reducing expressiveness.
>
> *What to seek:* Higher isotropy generally indicates better-quality embeddings. Values > 0.1 are reasonable; > 0.3 is good. Lower-dimensional embeddings tend to have higher isotropy.

**Average Norm**
> *Definition:* Mean magnitude (L2 norm) of word vectors in the embedding space.
>
> *Intuition:* Indicates the typical "length" of vectors. Consistent norms suggest stable training; high variance may indicate some words are undertrained.
>
> *What to seek:* Relatively consistent norms across models. The absolute value matters less than consistency (low std deviation).

**Cosine Similarity**
> *Definition:* Measures angular similarity between vectors, ranging from -1 (opposite) to 1 (identical direction).
>
> *Intuition:* Words with similar meanings should have high cosine similarity. This is the standard metric for semantic relatedness in embeddings.
>
> *What to seek:* Semantically related words should score > 0.5; unrelated words should be near 0. Synonyms often score > 0.7.

**t-SNE Visualization**
> *Definition:* t-Distributed Stochastic Neighbor Embedding - a dimensionality reduction technique that preserves local structure for visualization.
>
> *Intuition:* Clusters in t-SNE plots indicate groups of semantically related words. Spread indicates vocabulary diversity; tight clusters suggest semantic coherence.
>
> *What to seek:* Meaningful clusters (e.g., numbers together, verbs together). Avoid over-interpreting distances - t-SNE preserves local, not global, structure.

### General Interpretation Guidelines

1. **Compare within model families:** Metrics are most meaningful when comparing models of the same type (e.g., 8k vs 64k tokenizer).
2. **Consider trade-offs:** Better performance on one metric often comes at the cost of another (e.g., compression vs. OOV rate).
3. **Context matters:** Optimal values depend on downstream tasks. Text generation may prioritize different metrics than classification.
4. **Corpus influence:** All metrics are influenced by corpus characteristics. Wikipedia text differs from social media or literature.
5. **Language-specific patterns:** Morphologically rich languages (like Arabic) may show different optimal ranges than analytic languages.


### Visualizations Index

| Visualization | Description |
|---------------|-------------|
| Tokenizer Compression | Compression ratios by vocabulary size |
| Tokenizer Fertility | Average token length by vocabulary |
| Tokenizer OOV | Unknown token rates |
| Tokenizer Total Tokens | Total tokens by vocabulary |
| N-gram Perplexity | Perplexity by n-gram size |
| N-gram Entropy | Entropy by n-gram size |
| N-gram Coverage | Top pattern coverage |
| N-gram Unique | Unique n-gram counts |
| Markov Entropy | Entropy by context size |
| Markov Branching | Branching factor by context |
| Markov Contexts | Unique context counts |
| Zipf's Law | Frequency-rank distribution with fit |
| Vocab Frequency | Word frequency distribution |
| Top 20 Words | Most frequent words |
| Vocab Coverage | Cumulative coverage curve |
| Embedding Isotropy | Vector space uniformity |
| Embedding Norms | Vector magnitude distribution |
| Embedding Similarity | Word similarity heatmap |
| Nearest Neighbors | Similar words for key terms |
| t-SNE Words | 2D word embedding visualization |
| t-SNE Sentences | 2D sentence embedding visualization |
| Position Encoding | Encoding method comparison |
| Model Sizes | Storage requirements |
| Performance Dashboard | Comprehensive performance overview |

---
## About This Project

### Data Source

Models trained on [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly) - a monthly snapshot of Wikipedia articles across 300+ languages.

### Project

A project by **[Wikilangs](https://wikilangs.org)** - Open-source NLP models for every Wikipedia language.

### Maintainer

[Omar Kamali](https://omarkamali.com) - [Omneity Labs](https://omneitylabs.com)

### Citation

If you use these models in your research, please cite:

```bibtex
@misc{wikilangs2025,
  author = {Kamali, Omar},
  title = {Wikilangs: Open NLP Models for Wikipedia Languages},
  year = {2025},
  doi = {10.5281/zenodo.18073153},
  publisher = {Zenodo},
  url = {https://huggingface.co/wikilangs}
  institution = {Omneity Labs}
}
```

### License

MIT License - Free for academic and commercial use.

### Links

- 🌐 Website: [wikilangs.org](https://wikilangs.org)
- 🤗 Models: [huggingface.co/wikilangs](https://huggingface.co/wikilangs)
- 📊 Data: [wikipedia-monthly](https://huggingface.co/datasets/omarkamali/wikipedia-monthly)
- 👤 Author: [Omar Kamali](https://huggingface.co/omarkamali)
- 🤝 Sponsor: [Featherless AI](https://featherless.ai)
---
*Generated by Wikilangs Models Pipeline*

*Report Date: 2026-01-11 02:36:16*