id string | title string | topic string | category string | language string | text string | segmented_text string | word_count int64 | token_count int64 | zwsp_count int64 | author string | is_professionally_edited bool | editorial_status string | edited_by string | edited_at string | editorial_notes string | review_checklist dict | tags list | rag_chunks list | summary string | created_at string |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
corpus-seed-khmer-llm | ការបង្កើតគំរូភាសាខ្មែរធំៗ (Khmer LLM) និងស្ថាបត្យកម្ម BiGRU+CRF សម្រាប់ Word Segmentation | Khmer LLM Architecture & Neural Word Segmentation | AI & Khmer NLP | km | # ការបង្កើតគំរូភាសាខ្មែរធំៗ (Khmer LLM) និងស្ថាបត្យកម្ម BiGRU+CRF សម្រាប់ Word Segmentation
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** អត្ថបទស្រាវជ្រាវបច្ចេកទេសនេះពិពណ៌នាយ៉ាងលម្អិតអំពីយុទ្ធសាស្ត្រនៃការបង្កើត Large Language Models (LLMs) សម្រាប់ភាស... | # ការបង្កើតគំរូភាសាខ្មែរធំៗ (Khmer LLM) និងស្ថាបត្យកម្ម BiGRU+CRF សម្រាប់ Word Segmentation
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** អត្ថបទស្រាវជ្រាវបច្ចេកទេសនេះពិពណ៌នាយ៉ាងលម្អិតអំពីយុទ្ធសាស្ត្រនៃការបង្កើត Large Language Models (LLMs) សម្រាប់ភាស... | 4,200 | 5,670 | 1,420 | Joseph Van Guan | true | professionally_edited | Joseph Van Guan | 2026-09-09T22:42:53.543Z | Peer-reviewed & verified for Hugging Face corpus. | {
"factChecked": true,
"nlpVerified": true,
"academicKhmer": true,
"schemaValid": true,
"hallucinationFree": true
} | [
"Khmer NLP",
"LLM",
"BiGRU+CRF",
"Word Segmentation",
"ZWSP",
"Tokenization"
] | [
{
"id": "chunk-llm-1",
"content": "បញ្ហា Khmer Tokenization គឺគ្មានការដកឃ្លារវាងពាក្យ (Scriptio Continua) នាំឱ្យ Tokenizer បរទេសបំបែកតួអក្សរខ្មែរជា byte fragments ច្រើនលើសលប់។",
"tokens": 45
},
{
"id": "chunk-llm-2",
"content": "ដំណោះស្រាយ BiGRU+CRF ប្រើប្រាស់ Character Embedding រួមជាមួយ Co... | ការស្រាវជ្រាវស៊ីជម្រៅលើស្ថាបត្យកម្ម Khmer LLM និងការកាត់ពាក្យ BiGRU+CRF ដើម្បីពង្រឹងប្រសិទ្ធភាព Tokenizer ភាសាខ្មែរ។ | 2026-09-09T22:42:53.543Z |
corpus-seed-bakong-fintech | ប្រព័ន្ធបាគង និងស្តង់ដារ KHQR៖ យុទ្ធសាស្ត្របរិវត្តកម្មហិរញ្ញវត្ថុ និងការទូទាត់ឌីជីថលឆ្លងដែននៅកម្ពុជា | Project Bakong, KHQR Standard & Cross-Border FinTech | FinTech & Bakong | km | # ប្រព័ន្ធបាគង និងស្តង់ដារ KHQR៖ យុទ្ធសាស្ត្របរិវត្តកម្មហិរញ្ញវត្ថុ និងការទូទាត់ឌីជីថលឆ្លងដែននៅកម្ពុជា
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** ប្រព័ន្ធបាគង (Project Bakong) ដែលត្រូវបានផ្តួចផ្តើមដោយធនាគារជាតិនៃកម្ពុជា (NBC) គឺជាប្រព័ន្ធទូទាត់កម្រិតជាតិផ្អែក... | # ប្រព័ន្ធបាគង និងស្តង់ដារ KHQR៖ យុទ្ធសាស្ត្របរិវត្តកម្មហិរញ្ញវត្ថុ និងការទូទាត់ឌីជីថលឆ្លងដែននៅកម្ពុជា
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** ប្រព័ន្ធបាគង (Project Bakong) ដែលត្រូវបានផ្តួចផ្តើមដោយធនាគារជាតិនៃកម្ពុជា (NBC) គឺជាប្រព័ន្ធទូទាត់កម្រិតជាតិផ្អែក... | 202 | 285 | 60 | Joseph Van Guan | true | professionally_edited | Joseph Van Guan | 2026-09-14T20:33:41.897Z | Professionally edited, verified against Royal Academy Khmer standard, and certified for Hugging Face. | {
"factChecked": true,
"nlpVerified": true,
"academicKhmer": true,
"schemaValid": true,
"hallucinationFree": true
} | [
"Project Bakong",
"KHQR",
"Blockchain",
"Hyperledger Iroha",
"FinTech",
"NBC",
"EMVCo"
] | [
{
"id": "chunk-bakong-1",
"content": "ប្រព័ន្ធបាគងប្រើប្រាស់ Hyperledger Iroha ជាមួយក្បួនដោះស្រាយ Sumeragi Consensus គាំទ្រទាំង KHR និង USD ដោយគ្មានកម្រៃសេវាផ្ទេរប្រាក់ឆ្លងធនាគារ។",
"tokens": 48
},
{
"id": "chunk-bakong-2",
"content": "ស្តង់ដារ KHQR អនុលោមតាម EMVCo គាំទ្រការទូទាត់ឆ្លងដែនជាមួ... | ការវិភាគបច្ចេកទេសលើប្រព័ន្ធបាគង ស្តង់ដារ KHQR EMVCo និងយន្តការទូទាត់ឌីជីថលឆ្លងដែនកម្ពុជា។ | 2026-09-10T22:42:53.544Z |
corpus-seed-cyber-cii | ការការពារហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (CII) និងច្បាប់សន្តិសុខសាយប័រនៅកម្ពុជា | Critical Information Infrastructure (CII) Protection & Cybersecurity Governance | Cybersecurity & Cloud | km | # ការការពារហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (CII) និងច្បាប់សន្តិសុខសាយប័រនៅកម្ពុជា
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** ហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (Critical Information Infrastructure - CII) គ្របដណ្តប់លើវិស័យធនាគារ ថាមពល ទូរគមនាគារ សុខាភិបាល និងរដ្ឋបាលសាធ... | # ការការពារហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (CII) និងច្បាប់សន្តិសុខសាយប័រនៅកម្ពុជា
> **សេចក្តីសង្ខេបប្រតិបត្តិ (Executive Abstract):** ហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (Critical Information Infrastructure - CII) គ្របដណ្តប់លើវិស័យធនាគារ ថាមពល ទូរគមនាគារ សុខាភិបាល និងរដ្ឋបាលសាធ... | 3,600 | 4,860 | 1,150 | Joseph Van Guan | true | professionally_edited | Joseph Van Guan | 2026-09-11T22:42:53.544Z | Peer-reviewed & verified for Hugging Face corpus. | {
"factChecked": true,
"nlpVerified": true,
"academicKhmer": true,
"schemaValid": true,
"hallucinationFree": true
} | [
"Cybersecurity",
"CII",
"NIST CSF",
"CamDX",
"ISO 27001",
"Ransomware Defense"
] | [
{
"id": "chunk-cyber-1",
"content": "CII នៅកម្ពុជាគ្របដណ្តប់លើ ៦ វិស័យសំខាន់រួមមាន ហិរញ្ញវត្ថុ ទូរគមនាគារ ថាមពល SCADA ដឹកជញ្ជូន សុខាភិបាល និងរដ្ឋបាលសាធារណៈ CamDX។",
"tokens": 46
}
] | ក្របខណ្ឌការពារហេដ្ឋារចនាសម្ព័ន្ធព័ត៌មានសំខាន់ៗ (CII) និងវិធានការសន្តិសុខសាយប័រសម្រាប់ស្ថាប័នរដ្ឋនិងឯកជន។ | 2026-09-11T22:42:53.544Z |
khmer-nlp-technical-corpus — Khmer Strategic NLP Corpus
Dataset Summary
This dataset contains peer-grade long-form technical treatises (3,000+ words each) in the Khmer language (km / ភាសាខ្មែរ). Every article is normalized and features neural BiGRU+CRF word segmentation with Zero-Width Space (\u200B) injection to prevent token fragmentation in sub-word tokenizers.
Dataset Statistics
Total Documents: 3
Train Documents: 3
Total Words: 8,002
Total Sub-Word Tokens: 10,815
ZWSP Count: 2,630 \u200B points
Loading with Python datasets
from datasets import load_dataset
# Load from Hugging Face Hub
ds = load_dataset("guanvireak/khmer-nlp-technical-corpus")
print(ds)
print(ds["train"][0]["title"])
Schema Definition
| Field | Type | Description |
|---|---|---|
id |
string | Unique document identifier |
title |
string | Formal article title in Khmer |
topic |
string | Core technical topic / research domain |
category |
string | Domain classification pillar |
language |
string | ISO 639-1 code (km) |
text |
string | Complete Markdown treatise (3,000+ words) |
segmented_text |
string | Text segmented with Zero-Width Space (\u200B) |
word_count |
int | Total words count |
token_count |
int | Estimated BPE tokens count |
zwsp_count |
int | ZWSP word boundary count |
tags |
list[string] | Keywords & domain entities |
rag_chunks |
list[dict] | Ready-to-index semantic chunks |
Generated via GV Draft Generator — Khmer Enhancement Layer.
- Downloads last month
- 43