sabin1234/Real_Scarp_ROMANIZED_Health_FAQ_Complile_Dataset

Romanized Nepali Health FAQ Dataset - Complete Documentation

Dataset Overview

The Merged Romanized Nepali Health FAQ Dataset is the romanized (Latin script) version of the comprehensive health information resource containing 3,174 question-answer pairs in Nepali language using romanized Nepali script. This dataset is ideal for applications that require Latin script input/output and training conversational AI systems for romanized Nepali text processing.

Key Statistics

Metric	Value
Total Records	3,174
Language	Nepali (Romanized)
Script	Latin/Roman Script (romanized transliteration)
License	Apache-2.0 (Permissive)
Data Format	JSONL (JSON Lines)
Task Type	Instruction-Following / Question-Answering
Generation Type	Real (Authentic)
Unique Health Domains	3 Major + 20+ Sub-domains
Unique Sources	20+ Medical Organizations & Healthcare Providers
Average Question Length	7.7 words
Average Answer Length	39.9 words
Metadata Script	Devanagari (नेपाली)

What Makes This Version Special

✅ Romanized Transcription: All question-answer pairs are written in romanized Nepali (Latin script)
✅ Consistent with Devanagari Version: Same content as the Devanagari version for easy comparison
✅ Standard Romanization: Uses common Nepali romanization conventions
✅ Preserved Metadata: All metadata remains in original Devanagari script
✅ Multilingual Processing: Perfect for NLP systems that handle romanized input

Example Comparison:

Aspect	Devanagari	Romanized
Script Type	नेपाली	Latin/Roman
Example Q	तौल व्यवस्थापनका लागि आधारभूत चयापचय दर किन महत्त्वपूर्ण छ?	taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?
Metadata	Devanagari	Devanagari
Content	Identical	Identical (romanized)

---

1. BEHAVIOR DIVERSITY ANALYSIS

The dataset includes diverse response behaviors designed to handle different types of health inquiries with varying levels of detail and formality.

Behavior Distribution

Behavior Type (Devanagari)	English Translation	Count	Percentage	Description
जानकारीमूलक उत्तर दिने	Informative Answer	1,458	45.9%	Provides comprehensive, factual health information translated to romanized Nepali
सामान्य जानकारी दिने	General Information	1,363	42.9%	Delivers general health knowledge without excessive technical details
छोटो उत्तर माग्ने	Brief Answer Request	215	6.8%	Concise responses addressing specific health queries
प्रश्नको प्रत्यक्ष उत्तर दिने	Direct Answer	100	3.2%	Direct, straightforward responses to health questions
छोटो र जानकारीमूलक उत्तर दिने	Short & Informative	38	1.2%	Balanced approach combining brevity with informative content

Behavior Insights

1. Informative vs. General Split: Nearly 89% of responses adopt an informative or general knowledge approach
2. Romanized Format: All responses are presented in romanized Nepali for Latin script compatibility
3. Response Conciseness: About 7% use brief responses for straightforward health queries
4. Direct Answers: Only 3.2% use purely direct answers, prioritizing comprehensive information

Example Romanized Behaviors

Informative Answer:

Q: taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?
A: aadharabhut chayapachaya dar thaha paunda sharirlai samanya rupma kam garna 
   aawashyak parne aadharabhut kyaloriko sankhya bujhna maddat hunchha...

Brief Answer:

Q: urwar samayawadhi bhaneko ke ho?
A: urwar samayawadhile mahinawari chakrabhitra garbhadharan hune sambhawana 
   sabaibhanda badhi hune dinharulai janaunchha.

---

2. QUESTION PATTERN DIVERSITY ANALYSIS

Question Type Distribution

Question Type (Devanagari)	English Translation	Count	Percentage	Romanized Example
तथ्यमा आधारित प्रश्न	Fact-Based Questions	2,831	89.2%	paragakanko star bhaneko ke ho?
प्रश्नोत्तर प्रश्न	Interrogative Questions	101	3.2%	Verification-seeking questions
स्वास्थ्यसम्बन्धी प्रश्न	Health-Specific Questions	100	3.2%	Domain-specific medical questions
प्रश्नोत्तर	Q&A Format	100	3.2%	Pre-formatted question-answer pairs
सामान्य प्रश्न	General Questions	4	0.1%	Broad health inquiries
N/A	Unclassified	38	1.2%	Uncategorized entries

Question Pattern Characteristics

Fact-Based Questions (89.2%)

• Pattern: "What is...?", "Why does...?", "How does...?" in romanized Nepali
• Romanized Examples:
  • paragakanko star bhaneko ke ho? (What is pollen count?)
  • samayasangai mero aadharabhut chayapachaya dar pariwartana huna sakchha? (Can my BMR change over time?)
  • ke ghansako paragbata hune jwaro ra dipresanabich sambandha chha? (Is there a link between hay fever and depression?)

Question Complexity Metrics

Metric	Value	Range
Average Word Count	7.7 words	2-28 words
Minimum Length	2 words	Simple queries
Maximum Length	28 words	Complex multi-part questions
Median Length	~6 words	Typical query

Romanized Question Examples by Complexity

Very Short (2-4 words):

• ke asthema ghatak hunchha? (Is asthma fatal?)
• kolon kansar ke ho? (What is colon cancer?)

Short (5-8 words):

• mero khanapanale aadharabhut chayapachaya darlai asar garcha? (Does my diet affect basal metabolic rate?)
• urwar samayawadhi bhaneko ke ho? (What is the fertile window?)

Medium (9-14 words):

• tapaiko wyaktigat swasthya awasthale dainik jiwanma kasari asar garcha bhanne dekhaune praman? (Evidence of how your health condition affects daily life?)

Complex (15+ words):

• tapaika lagi ghansako paragbata hune jwaroma prayog garne kuna prakarko nakko chharkai upayukta hunchha? (What type of nasal spray is suitable for hay fever treatment for you?)

---

3. DOMAIN AND CATEGORY DIVERSITY

Primary Domain Distribution

Domain (Devanagari)	English	Count	Percentage	Description
स्वास्थ्य	Health	3,036	95.7%	Core health and medical topics (romanized)
एस्थेटिक मेडिसिन	Aesthetic Medicine	100	3.2%	Cosmetic and aesthetic procedures
शिक्षा	Education	38	1.2%	Educational/academic health topics

Top Categories (Top 15 with Romanized Examples)

Category (Devanagari)	Count	%	Romanized Topic
स्वास्थ्य सम्बन्धी प्रश्नोत्तर	1,235	38.9%	Health Q&A - RadFAQs content
स्वास्थ्यसम्बन्धी प्रायः सोधिने प्रश्न	938	29.6%	Frequently asked health questions - Patient.info
क्यान्सर सम्बन्धी प्रायः सोधिने प्रश्न	191	6.0%	Cancer FAQs - Colorectal cancer focus
एस्थेटिक अस्पताल प्रश्नोत्तर	101	3.2%	Cosmetic procedures - Healthy Choice
छाती तथा एलर्जीसम्बन्धी स्वास्थ्य प्रश्न	100	3.2%	dam tatha elarji - Respiratory & allergies
स्वास्थ्य, फिटनेस र पोषण	100	3.2%	Health, fitness & nutrition topics
भिटिलिगो तथा छालाको स्वास्थ्य	99	3.1%	vitiligo - Dermatological conditions
अज्ञात	99	3.1%	Unknown categories
राष्ट्रिय स्वास्थ्य सेवा जागिर	97	3.1%	NHS employment & careers
गर्भनिरोधक सम्बन्धी प्रायः सोधिने प्रश्न	53	1.7%	Contraception FAQs

Sub-Domain Examples (Romanized Mapping)

Medical Specialty	Romanized Term Examples	Count	%
Radiology & Imaging	RadFAQs content	1,235	38.9%
General Health	swasthya sambandhi prashna	496	15.6%
Oncology	kansar topics	191	6.0%
Respiratory	dam tatha elarji (Asthma & Allergy)	100	3.2%
Dermatology	vitiligo, chhalako swasthya	99	3.1%
Reproductive Health	garbhnirodak, urwar samayawadhi	53	1.7%

---

4. SOURCE AND ORGANIZATION DIVERSITY

Source Distribution (Top 20)

Source Organization	Romanized Reference	Count	%	Focus
RadFAQs	Radiology FAQs	1,235	38.9%	Medical imaging Q&A
Patient.info	Patient information portal	938	29.6%	General patient information
Cancer Clinic Pune	Oncology clinic content	191	6.0%	Cancer treatment
CohereLabs/aya_dataset	Multilingual health data	101	3.2%	Research dataset
Chest Clinic of Kathmandu & Allergy Center	dam tatha elarji kendra	100	3.2%	Respiratory & allergies
Harley Street Institute	Medical institute	100	3.2%	General medicine
SAC Nepal	Nepal health provider	100	3.2%	Nepalese services
Fight Vitiligo	vitiligo niyantran organization	99	3.1%	Skin conditions
Global Healthcare Jobs	Healthcare careers	97	3.1%	Healthcare employment
Nepal CRS Company	Nepal healthcare company	53	1.7%	Reproductive health
Future Education	MBBS in Nepal provider	38	1.2%	Medical education
स्वास्थ्य बीमा बोर्ड	Health Insurance Board	29	0.9%	Insurance info
राष्ट्रिय रक्तसञ्चार सेवा	National Blood Service	26	0.8%	Blood donation
Mechi Eye Hospital	Eye care center	13	0.4%	Ophthalmology

Geographic Coverage

Region	Sources	Records	Focus
Nepal	Chest Clinic, SAC Nepal, Eye Hospital, National Services	600+	Local health services
India	Cancer Clinic Pune	191	Oncology expertise
United Kingdom	Patient.info, RadFAQs, Harley Street	2,100+	Comprehensive health info
International	WHO, CohereLabs, Research datasets	183+	Multilingual resources

---

5. TEXT LENGTH AND ROMANIZATION ANALYSIS

Question Length Distribution

Metric	Value	Range	Romanized Observation
Average Length	7.7 words	2-28 words	Similar to Devanagari version
Minimum Length	2 words	Simple queries	ke ho? type questions
Maximum Length	28 words	Complex queries	Romanization preserves word count
Median Length	~6 words	Typical query	Most queries remain brief

Answer Length Distribution

Metric	Value	Range	Description
Average Length	39.9 words	1-234 words	Comprehensive responses
Minimum Length	1 word	Single-word answers	Rare minimal responses
Maximum Length	234 words	Detailed explanations	Maximum detail answers
Median Length	~35 words	Typical answer	Standard informative response

Romanized Text Examples

Short Response Example:

Q: urwar samayawadhi bhaneko ke ho?
(What is the fertile window?)

A: urwar samayawadhile mahinawari chakrabhitra garbhadharan hune sambhawana 
   sabaibhanda badhi hune dinharulai janaunchha. andotsarjana ganana garne 
   sadhana yahi wishesh samayawadhiko anumana garna banaieka hunchhan.
(34 words)

Medium Response Example:

Q: mero khanapanale aadharabhut chayapachaya darlai asar garcha?
(Does my diet affect my basal metabolic rate?)

A: lekhma khanapanale aadharabhut chayapachaya darlai kasari asar garcha bhanne 
   kura pratyaksha rupma ullekh garieko chhaina, tara aadharabhut chayapachaya 
   dar bhaneko sharirik gatiwidhi nagare pani sharirle aadharabhut kamka lagi 
   prayog garne kyaloriko sankhya ho. yasko prabhav...
(~50 words)

Romanization Characteristics

Aspect	Details
Vowels	a, e, i, o, u used consistently
Consonants	Standard romanization conventions applied
Special Characters	"aa", "ch", "sh", "kh" for Nepali sounds
Word Boundaries	Spaces preserved from original
Apostrophes	Not used; romanization is phonetic
Diacritics	None used (pure Latin script)

Romanization Examples:

Devanagari	Romanized	English
तौल	taul	weight
चयापचय	chayapachaya	metabolism
आधारभूत	aadharabhut	basal
स्वास्थ्य	swasthya	health
महत्त्वपूर्ण	mahattwapurna	important
परागकण	paragakan	pollen
दर	dar	rate

---

6. HOW TO LOAD AND USE THE ROMANIZED DATASET

6.1 Basic Loading

import json

def load_romanized_nepali_health_faq():
    """Load romanized Nepali Health FAQ dataset."""
    dataset = []
    
    with open('merged_romanized_health_faq_dataset.jsonl', 'r', encoding='utf-8') as f:
        for line in f:
            dataset.append(json.loads(line))
    
    print(f"✓ Loaded {len(dataset)} records in romanized format")
    return dataset

# Load data
data = load_romanized_nepali_health_faq()

# Access romanized content
first = data[0]
print(f"ID: {first['id']}")
print(f"Question (Romanized): {first['conversations'][0]['value']}")
print(f"Answer (Romanized): {first['conversations'][1]['value']}")
print(f"Metadata (Devanagari): {first['metadata_json']}")

Output:

✓ Loaded 3174 records in romanized format
ID: nepali_health_faq_00001
Question (Romanized): taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?
Answer (Romanized): aadharabhut chayapachaya dar thaha paunda sharirlai samanya rupma kam garna...
Metadata (Devanagari): {"generation_domain": "स्वास्थ्य", ...}

6.2 Extract Romanized Q&A Pairs

import json

def extract_romanized_qa_pairs(filepath):
    """Extract romanized Q&A pairs."""
    qa_pairs = []
    
    with open(filepath, 'r', encoding='utf-8') as f:
        for record in f:
            data = json.loads(record)
            conversations = data.get('conversations', [])
            
            for i, conv in enumerate(conversations):
                if conv['from'] == 'human' and i+1 < len(conversations):
                    if conversations[i+1]['from'] == 'gpt':
                        qa_pairs.append({
                            'question': conv['value'],  # Romanized
                            'answer': conversations[i+1]['value'],  # Romanized
                            'source': data.get('source'),
                            'id': data.get('id')
                        })
    
    return qa_pairs

# Usage
pairs = extract_romanized_qa_pairs('merged_romanized_health_faq_dataset.jsonl')
print(f"✓ Extracted {len(pairs)} romanized Q&A pairs")

# Display romanized example
print(f"\nRomanized Question:\n{pairs[0]['question']}")
print(f"\nRomanized Answer:\n{pairs[0]['answer']}")

6.3 Convert Romanized to Devanagari (Optional)

# For applications needing bidirectional conversion, you can use:
# 1. Google's Transliteration API
# 2. The nptxt library: pip install nptxt
# 3. Custom mapping functions

from nptxt import nepliteral

def romanized_to_devanagari(romanized_text):
    """Convert romanized Nepali to Devanagari."""
    return nepliteral.to_nepali(romanized_text)

# Example
romanized_q = "taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?"
devanagari_q = romanized_to_devanagari(romanized_q)
print(f"Romanized: {romanized_q}")
print(f"Devanagari: {devanagari_q}")

6.4 Using Advanced Loader with Romanized Data

# Use the same dataset_loader_complete.py class
from dataset_loader_complete import NepaliHealthFAQDataset

# Initialize with romanized dataset
dataset = NepaliHealthFAQDataset('merged_romanized_health_faq_dataset.jsonl')
records = dataset.load(verbose=False)

# All romanized content is accessible
qa = dataset.get_qa_pair(0)
print(f"Romanized Question: {qa['question']}")
print(f"Romanized Answer: {qa['answer']}")
print(f"Source: {qa['source']}")

# Search in romanized content
health_q = dataset.search_questions('swasthya')
print(f"Questions containing 'swasthya': {len(health_q)}")

# Export romanized data
dataset.export_to_csv('romanized_health_faq.csv')
print("✓ Exported romanized data to CSV")

6.5 Pandas DataFrame with Romanized Text

import pandas as pd
from dataset_loader_complete import NepaliHealthFAQDataset

# Load and convert
dataset = NepaliHealthFAQDataset('merged_romanized_health_faq_dataset.jsonl')
dataset.load(verbose=False)
df = dataset.to_dataframe()

# Verify romanized content
print("Sample romanized questions:")
print(df['question'].head())

---

7. DATASET COMPOSITION SUMMARY

Complete Statistical Overview

Aspect	Value	Count	Percentage
Total Records	-	3,174	100%
Script Format	Romanized (Latin)	3,174	100%
Metadata Script	Devanagari	3,174	100%
Behavior Types	5 distinct types	-	-
	Informative	1,458	45.9%
	General Info	1,363	42.9%
	Brief Answer	215	6.8%
	Direct Answer	100	3.2%
	Mixed Approach	38	1.2%
Question Types	6 types	-	-
	Fact-Based	2,831	89.2%
	Interrogative	101	3.2%
	Health-Specific	100	3.2%
	Structured Q&A	100	3.2%
	General	4	0.1%
	Unclassified	38	1.2%
Text Metrics	Avg Question	7.7	words
	Avg Answer	39.9	words
	Max Question	28	words
	Max Answer	234	words

---

8. UNIQUE FEATURES OF ROMANIZED VERSION

Advantages

✅ Latin Script Compatibility: Works with systems that don't support Devanagari
✅ Searchable ASCII Text: Full-text search on romanized content
✅ Global Input Method: No special keyboard layouts needed
✅ API Friendly: Easy integration with REST APIs and web services
✅ Preserved Meaning: Identical content to Devanagari version
✅ Metadata Bilingual: Queries in romanized, metadata in Devanagari
✅ NLP Research: Perfect for romanized Nepali NLP studies

Use Cases

1. Romanized Nepali Processing: Train NLP models on romanized text
2. Web Applications: Display romanized Nepali in web interfaces
3. Mobile Apps: Support typing without special keyboards
4. Global Accessibility: Easier for international users
5. Transliteration Research: Study romanized ↔ Devanagari conversion
6. Multilingual Systems: Process romanized with other Latin-based languages
7. Social Media: Romanized Nepali common on Facebook/Twitter
8. Informal Communication: Mirrors how many write Nepali informally

---

9. ROMANIZATION GUIDE

Common Romanization Patterns

Devanagari	Romanized	Example Word	English
अ	a	aadhar	foundation
आ	aa	aapho	walnut
इ	i	tiho	thin
ई	ii	chii	fine
उ	u	guru	teacher
ऊ	uu	bhuu	earth
ए	e	desha	country
ऐ	ai	baitar	barrel
ओ	o	roma	Rome
औ	au	kaula	family
क	k	kam	work
ख	kh	khana	food
ग	g	ghar	house
घ	gh	ghata	clock
च	ch	chayo	went
छ	chh	chhatri	umbrella
ज	j	janu	knee
झ	jh	jhanda	flag
ट	t	talo	below
ठ	th	thapai	plate
ड	d	dada	elder brother
ढ	dh	dhaato	dentist
त	t	tara	star
थ	th	thali	plate
द	d	dur	far
ध	dh	dhulo	dust
न	n	nala	lake
प	p	pani	water
फ	ph	pharpha	flutter
ब	b	balaak	child
भ	bh	bhanda	than
म	m	maan	dignity
य	y	yaad	memory
र	r	raat	night
ल	l	lanko	rope
व	w/v	wakta	time
श	sh	shanti	peace
ष	sh	shakha	branch
स	s	sano	small
ह	h	haat	hand

Example Conversions

Question:

Devanagari:  तौल व्यवस्थापनका लागि आधारभूत चयापचय दर किन महत्त्वपूर्ण छ?
Romanized:   taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?
English:     Why is basal metabolic rate important for weight management?

Answer:

Devanagari:  आधारभूत चयापचय दर थाहा पाउँदा शरीरलाई सामान्य रूपमा काम गर्न आवश्यक पर्ने...
Romanized:   aadharabhut chayapachaya dar thaha paunda sharirlai samanya rupma kam garna aawashyak parne...
English:     Understanding basal metabolic rate helps you know the basic calories needed for your body...

---

10. COMPARISON: ROMANIZED vs DEVANAGARI

Side-by-Side Examples

Aspect	Devanagari	Romanized
Script	नेपाली	Latin/Roman
Example Q1	परागकणको स्तर भनेको के हो?	paragakanko star bhaneko ke ho?
Example A1	परागकणको स्तरले कुनै निश्चित समयमा हावामा रहेको परागकणको मात्रालाई जनाउँछ।	paragakanko starle kunai nishchit samayama hawama raheko paragakanko matralai janaunchha.
File Size	~120 MB	~100 MB
Readability	Native Nepali	International audience
Keyboard Required	Devanagari input	Standard QWERTY
Search Capability	Script-specific	Standard ASCII search
Metadata	Devanagari	Devanagari
Content	Identical meaning	Identical meaning

---

11. RECOMMENDED USE CASES

Perfect For:

1. NLP Training: Train romanized Nepali language models
2. Web Platforms: Display romanized in web/mobile interfaces
3. Romanized Text Search: Build searchable FAQ systems
4. Transliteration Tools: Develop romanization/conversion utilities
5. Global Health Portals: Serve international users accessing Nepali content
6. Social Media Analysis: Analyze romanized Nepali health discussions
7. Typing Predictors: Build autocomplete for romanized input
8. Language Learning: Teach Nepali using romanized text
9. API Services: Build REST APIs for romanized content
10. Data Processing: ETL pipelines for romanized data

Not Ideal For:

❌ Native Nepali desktop applications
❌ Print publications in Nepali
❌ Schools teaching Devanagari script
❌ Traditional Nepali content libraries

---

12. MIGRATION AND COMPARISON WITH DEVANAGARI

Using Both Versions Together

import json

def load_both_versions():
    """Load and compare both romanized and Devanagari versions."""
    
    romanized_data = []
    with open('merged_romanized_health_faq_dataset.jsonl', 'r', encoding='utf-8') as f:
        for line in f:
            romanized_data.append(json.loads(line))
    
    devanagari_data = []
    with open('merged_nepali_health_faq_dataset.jsonl', 'r', encoding='utf-8') as f:
        for line in f:
            devanagari_data.append(json.loads(line))
    
    # Compare records
    for i in range(min(3, len(romanized_data))):
        rom = romanized_data[i]
        dev = devanagari_data[i]
        
        print(f"\n{'='*80}")
        print(f"Record {i+1}:")
        print(f"{'='*80}")
        print(f"\nRomanized Q: {rom['conversations'][0]['value']}")
        print(f"Devanagari Q: {dev['conversations'][0]['value']}")
        print(f"\nRomanized A: {rom['conversations'][1]['value'][:100]}...")
        print(f"Devanagari A: {dev['conversations'][1]['value'][:100]}...")
        print(f"\nSame Metadata: {rom['metadata_json'] == dev['metadata_json']}")

# Run comparison
load_both_versions()

Script Conversion Function

def detect_script(text):
    """Detect if text is Devanagari or Romanized."""
    # Devanagari script range
    devanagari_chars = set(range(0x0900, 0x0950))
    
    for char in text:
        if ord(char) in devanagari_chars:
            return "Devanagari"
    
    return "Romanized"

# Test
q1 = "तौल व्यवस्थापनका लागि..."
q2 = "taul wyawasthapanaka lagi..."

print(detect_script(q1))  # Devanagari
print(detect_script(q2))  # Romanized

---

13. TECHNICAL SPECIFICATIONS

File Format Details

• Format: JSONL (JSON Lines)
• Encoding: UTF-8 with romanized Latin script
• File Size: ~100 MB (estimated)
• Records per Line: 1 complete JSON object
• Compression: Recommended for storage (gzip)

Data Structure

{
  "id": "nepali_health_faq_00001",
  "conversations": [
    {
      "from": "human",
      "value": "taul wyawasthapanaka lagi aadharabhut chayapachaya dar kina mahattwapurna chha?"
    },
    {
      "from": "gpt",
      "value": "aadharabhut chayapachaya dar thaha paunda sharirlai samanya rupma kam garna aawashyak parne aadharabhut kyaloriko sankhya bujhna maddat hunchha..."
    }
  ],
  "source": "Patient.info",
  "language": "ne",
  "language_code": "npi",
  "script": "Deva",
  "license": "Apache-2.0",
  "metadata_json": "{...Devanagari metadata...}"
}

Key Fields

• id: Unique record identifier
• conversations: Array of [human question, gpt answer] pairs
• value: Romanized text content (Latin script)
• source: Original healthcare provider
• metadata_json: Metadata in Devanagari (नेपाली)
• All other fields: Same as Devanagari version

---

14. QUICK REFERENCE TABLES

Romanized Behaviors

Behavior	Count	%	Romanized Label
Informative Answer	1,458	45.9%	jankarimulak uttar dine
General Information	1,363	42.9%	samanya jankari dine
Brief Answer	215	6.8%	choto uttar mag_ne
Direct Answer	100	3.2%	prashnako pratyaksha uttar
Mixed	38	1.2%	mixed approach

Romanized Question Types

Type	Count	%	Romanized Example
Fact-Based	2,831	89.2%	ke ho? kina ho? kasari ho?
Interrogative	101	3.2%	Verification questions
Health-Specific	100	3.2%	Domain specific queries
Q&A Format	100	3.2%	Pre-formatted pairs

---

15. GETTING STARTED CHECKLIST

• Download merged_romanized_health_faq_dataset.jsonl
• Verify UTF-8 encoding (supports romanized text)
• Install required libraries (json, pandas)
• Run basic load test
• Extract Q&A pairs
• Generate statistics
• Export to preferred format
• Test romanized content handling
• (Optional) Compare with Devanagari version

---

16. SUPPORT AND CITATION

Citation

@dataset{romanized_nepali_health_faq_2024,
  title={Merged Romanized Nepali Health FAQ Dataset},
  author={Multiple Healthcare Organizations},
  year={2024},
  license={Apache-2.0},
  language={Nepali (Romanized)},
  script={Latin/Roman},
  records={3174},
  note={Romanized version of the Nepali Health FAQ Dataset}
}

Acknowledgment

This romanized version maintains identical content as the Devanagari version. Acknowledge both:

• Original healthcare organizations (as listed in source list)
• Romanization conversion process

---

Dataset Version: v1.0 (Romanized)
Base Version: v1.0 (Devanagari)
Total Records: 3,174 Q&A pairs
Script: Romanized (Latin/Roman)
License: Apache-2.0 (Permissive)
Last Updated: 2024

---

Quick Navigation

• README.md - Devanagari version documentation
• QUICKSTART.md - Quick examples and common tasks
• dataset_loader_complete.py - Python loader class
• README_ROMANIZED.md - This file (Romanized version)