This is Part 6 of the “Why Hotel API Integration Is So Hard?” series.

Part 1: Authentication Hell | Part 2: Data Chaos | Part 3: Rate Limiting Nightmare | Part 4: Error Handling | Part 5: Timezone Issues | Part 7: Summary (Coming Soon)


Have You Seen the Same Room with 5 Different Names?

Real-World Scenario

User searches for “King Room”:

Supplier A returns:

{
  "room_id": "R101",
  "room_name": "King Room",
  "bed_type": "King",
  "area": "25 m²",
  "amenities": ["WiFi", "TV", "Air conditioning"]
}

Supplier B returns:

{
  "room_id": "DELUXE_KING",
  "room_name": "Deluxe King - City View",
  "bed_type": "King Bed",
  "area": "28 m²",
  "amenities": ["WiFi", "TV", "Air Conditioning", "City View"]
}

Supplier C returns:

{
  "room_id": "KING_CITY",
  "room_name": "King Bed City View",
  "bed_type": "KING",
  "area": "26 m²",
  "amenities": ["Free WiFi", "Flat-screen TV", "Air Conditioning", "City View"]
}

Supplier D returns:

{
  "room_id": "101",
  "room_name": "King City View Room",
  "bed_type": "King",
  "area": "25 m²",
  "amenities": ["WiFi", "Television", "Air Conditioning"]
}

Supplier E returns:

{
  "room_id": "SUP_KING",
  "room_name": "Superior King Room",
  "bed_type": "King",
  "area": "27 m²",
  "amenities": ["Complimentary WiFi", "TV", "Climate Control", "City View"]
}

Here’s the Problem

Is this the same room?

  • All have King Bed
  • All have City View
  • Similar area (25-28 m²)
  • Similar amenities (WiFi, TV, Air Conditioning)

But:

  • Different room names (5 types)
  • Different room IDs (5 types)
  • Different amenity descriptions

How does the user choose?


Pain Point 1: String Matching Is Unreliable

Real-World Scenario

User searches for “King Room”

Your matching logic:

def match_rooms(user_query, supplier_rooms):
    matches = []
    for room in supplier_rooms:
        if user_query.lower() in room.name.lower():
            matches.append(room)
    return matches

Results:

User Query Room Name Match? Why?
King Room King Room ✅ Exact match
King Room Deluxe King - City View ❌ “King” is there, but “Deluxe” is not in query
King Room King Bed City View ❌ “King” is there, but “Bed” is not in query
King Room King City View Room ❌ “King” is there, but has extra “City View”
King Room Superior King Room ❌ “King” is there, but has extra “Superior”

Problem:

  • ❌ 5 suppliers’ King Rooms, only 1 matches
  • User only sees 1 option, should see 5

Improvement:

def match_rooms_v2(user_query, supplier_rooms):
    matches = []
    for room in supplier_rooms:
        # Extract keywords
        room_keywords = extract_keywords(room.name)
        user_keywords = extract_keywords(user_query)
        
        # Check if contains user keywords
        if set(user_keywords) <= set(room_keywords):
            matches.append(room)
    
    return matches

def extract_keywords(room_name):
    keywords = []
    # Extract bed type
    if "king" in room_name.lower():
        keywords.append("king")
    elif "queen" in room_name.lower():
        keywords.append("queen")
    elif "twin" in room_name.lower():
        keywords.append("twin")
    
    # Extract view
    if "city" in room_name.lower():
        keywords.append("city_view")
    elif "sea" in room_name.lower():
        keywords.append("sea_view")
    elif "garden" in room_name.lower():
        keywords.append("garden_view")
    
    # Extract class
    if "deluxe" in room_name.lower():
        keywords.append("deluxe")
    elif "superior" in room_name.lower():
        keywords.append("superior")
    
    return keywords

Results:

User Query Room Name Extracted Keywords Match?
King Room King Room [king] ✅
King Room Deluxe King - City View [king, city_view, deluxe] ❌ (extra deluxe and city_view)
King Room King Bed City View [king, city_view] ❌ (extra city_view)
King Room King City View Room [king, city_view] ❌ (extra city_view)
King Room Superior King Room [king, superior] ❌ (extra superior)

Problem:

  • ⚠️ Still only 1 supplier matches
  • ⚠️ Relies on rules, need to maintain many rules
  • ⚠️ Synonym issues: “City View” vs “Urban View” vs “Metropolis View”

Pain Point 2: Semantic Matching Is Difficult

Real-World Scenario

User searches for “Deluxe King Room”

Room types returned by suppliers:

Supplier Room Name Should Match?
A King Room ❌ (No Deluxe)
B Deluxe King - City View ✅
C King Bed City View ❌ (No Deluxe)
D King City View Room ❌ (No Deluxe)
E Superior King Room ⚠️ (Superior vs Deluxe, are they the same?)

Problem: Deluxe vs Superior vs Premium

Deluxe means:

  • Better bed
  • Larger area
  • More amenities

Superior means:

  • Better bed
  • Larger area
  • More amenities

Premium means:

  • Better bed
  • Larger area
  • More amenities

Problem:

  • Deluxe, Superior, Premium, are they the same?
  • Or different tiers?
  • If user searches for “Deluxe King”, should “Superior King” match?

Issues with Semantic Matching

Method 1: Synonym Mapping

synonym_mapping = {
    "deluxe": ["superior", "premium", "executive"],
    "superior": ["deluxe", "premium", "executive"],
    "premium": ["deluxe", "superior", "executive"],
}

def match_with_synonyms(user_query, supplier_rooms):
    user_keywords = extract_keywords(user_query)
    expanded_keywords = []
    for kw in user_keywords:
        if kw in synonym_mapping:
            expanded_keywords.extend(synonym_mapping[kw])
        else:
            expanded_keywords.append(kw)
    
    matches = []
    for room in supplier_rooms:
        room_keywords = extract_keywords(room.name)
        if set(expanded_keywords) & set(room_keywords):
            matches.append(room)
    
    return matches

Problem:

  • ⚠️ Relies on manually maintained synonym mapping
  • ⚠️ Synonyms may not be accurate enough
  • ⚠️ Different suppliers may have different synonyms

Method 2: Word Vector Similarity

from sentence_transformers import SentenceTransformer

model = SentenceTransformer('all-MiniLM-L6-v2')

def match_with_embeddings(user_query, supplier_rooms):
    # 1. Convert user query to vector
    user_embedding = model.encode(user_query)
    
    matches = []
    for room in supplier_rooms:
        # 2. Convert room name to vector
        room_embedding = model.encode(room.name)
        
        # 3. Calculate cosine similarity
        similarity = cosine_similarity(user_embedding, room_embedding)
        
        # 4. If similarity > threshold, match
        if similarity > 0.8:
            matches.append(room)
    
    return matches

Problem:

  • ⚠️ Requires training or loading pre-trained model
  • ⚠️ Requires computing resources
  • ⚠️ Similarity threshold needs tuning

Pain Point 3: Incomplete Attributes

Real-World Scenario

User searches for “King Room with City View”

Supplier A returns:

{
  "room_name": "King Room",
  "bed_type": "King",
  "area": "25 m²",
  "amenities": ["WiFi", "TV", "Air conditioning"]
}

Supplier B returns:

{
  "room_name": "Deluxe King - City View",
  "bed_type": "King",
  "area": "28 m²",
  "amenities": ["WiFi", "TV", "Air Conditioning", "City View"]
}

Supplier C returns:

{
  "room_name": "King Bed City View",
  "bed_type": "KING",
  "area": "26 m²",
  "amenities": ["Free WiFi", "Flat-screen TV", "Air Conditioning", "City View"]
}

Problem: Inconsistent Attributes

Supplier Has City View? How Represented?
A ❌ None
B ✅ In amenities
C ✅ In amenities
D ✅ In room_name
E ✅ In room_name

Problem:

  • Some suppliers put City View in amenities
  • Some suppliers put City View in room_name
  • Some suppliers have no City View attribute at all

User complaints:

  • “Why do some rooms have City View and some don’t?”
  • “I searched for King Room with City View, why do some results not have City View?”

Pain Point 4: Price Mismatch

Real-World Scenario

User books “Deluxe King - City View” ($180)

Order confirmation:

  • Room type: Superior King Room (Supplier E)
  • Price: $150

User complains:

  • “I booked Deluxe King - City View ($180), why does the order show Superior King Room ($150)?”
  • “The price is wrong too!”

Cause:

  • Your matching logic thinks “Deluxe King - City View” and “Superior King Room” are the same room type
  • Actually, these two room types are different (Deluxe vs Superior)
  • Prices are also different ($180 vs $150)

Problem: False Match

Matching logic:

# Simple string matching
if "King" in room_name:
    return "King Room"  # All King matched to "King Room"

Problem:

  • ❌ Deluxe King and Superior King both matched to “King Room”
  • ❌ User thinks they booked Deluxe King, actually booked Superior King
  • ❌ Price is wrong

Pain Point 5: Users Complain Rooms Don’t Match Photos

Real-World Scenario

User books “Deluxe King - City View”

Photo seen:

  • King Bed (Deluxe)
  • City View (large window)
  • 28 m² area

Actual check-in:

  • King Bed (standard)
  • City View (standard window)
  • 25 m² area

User complains:

  • “The photo shows Deluxe King, but I only got a standard King!”
  • “Photo shows 28 m², actually only 25 m²!”
  • “Photo shows large window, actually just a standard window!”

Cause:

  • Supplier A’s “Deluxe King” = Standard King Bed + City View
  • Supplier B’s “Deluxe King” = Deluxe King Bed + City View + 28 m²
  • Supplier C’s “Deluxe King” = King Bed + Deluxe Room + City View

Problem:

  • Different suppliers define “Deluxe” differently
  • Different suppliers define area differently (may include balcony)
  • Different suppliers define view differently (may have different levels)

How We Solve It

Core Approach: GIATA + AI Room Mapping

Problem: 5 suppliers, 5 different room type names, incomplete attributes, prone to false matching.

Solution: GIATA authoritative database + AI room mapping + attribute supplementing + consistency verification


1. GIATA Authoritative Database

What Is GIATA?

  • Global hotel information database
  • 500,000+ hotels
  • 1,000,000+ room types
  • Standardized room type names and attributes
  • Authoritative room mapping standard

GIATA Room Type Standards

GIATA Room Type Bed Type Area Amenities
STANDARD_KING King 20-25 m² Basic amenities
DELUXE_KING King 25-30 m² Basic + more
SUPERIOR_KING King 25-30 m² Basic + more
PREMIUM_KING King 30-35 m² Basic + most
EXECUTIVE_KING King 30-35 m² Basic + most

GIATA Query

import giata

giata_client = giata.Client(api_key="YOUR_GIATA_API_KEY")

def query_giata(hotel_id, supplier_room_name):
    # 1. Query GIATA database
    result = giata_client.search(
        hotel_id=hotel_id,
        room_name=supplier_room_name
    )
    
    # 2. If found, return GIATA standard room type
    if result:
        return {
            "giata_room_type": result.room_type,  # "DELUXE_KING"
            "standard_name": result.standard_name,  # "Deluxe King Room"
            "bed_type": result.bed_type,  # "King"
            "area_min": result.area_min,  # 25
            "area_max": result.area_max,  # 30
            "amenities": result.amenities  # ["WiFi", "TV", "AC", "City View"]
        }
    
    # 3. If not found, use AI matching
    return match_with_ai(hotel_id, supplier_room_name)

2. AI Room Mapping

Word Vector Similarity

from sentence_transformers import SentenceTransformer

model = SentenceTransformer('all-MiniLM-L6-v2')

def match_with_ai(hotel_id, supplier_room_name):
    # 1. Get all GIATA standard room types for this hotel
    giata_rooms = giata_client.get_standard_rooms(hotel_id)
    
    # 2. Convert supplier room type to vector
    supplier_embedding = model.encode(supplier_room_name)
    
    # 3. Calculate similarity with each GIATA standard room type
    similarities = []
    for giata_room in giata_rooms:
        giata_embedding = model.encode(giata_room.standard_name)
        similarity = cosine_similarity(supplier_embedding, giata_embedding)
        similarities.append((giata_room, similarity))
    
    # 4. Select GIATA room type with highest similarity
    sorted_similarities = sorted(similarities, key=lambda x: x[1], reverse=True)
    best_match = sorted_similarities[0]
    
    # 5. If similarity > threshold, return match result
    if best_match[1] > 0.85:
        return {
            "giata_room_type": best_match[0].room_type,
            "standard_name": best_match[0].standard_name,
            "confidence": best_match[1],  # 0.85-1.0
            "method": "ai_embedding"
        }
    
    # 6. If similarity not high enough, return low confidence result
    return {
        "giata_room_type": best_match[0].room_type,
        "standard_name": best_match[0].standard_name,
        "confidence": best_match[1],
        "method": "ai_embedding",
        "warning": "Low confidence, please review"
    }

Rule Engine (Auxiliary)

def match_with_rules(supplier_room):
    # 1. Extract room type features
    bed_type = extract_bed_type(supplier_room.name)
    view_type = extract_view_type(supplier_room.name)
    room_class = extract_room_class(supplier_room.name)  # Deluxe/Superior/Premium
    
    # 2. Match GIATA room type based on features
    if bed_type == "King" and view_type == "City View":
        if room_class == "Deluxe":
            return "DELUXE_KING_CITY_VIEW"
        elif room_class == "Superior":
            return "SUPERIOR_KING_CITY_VIEW"
        elif room_class == "Premium":
            return "PREMIUM_KING_CITY_VIEW"
        else:
            return "KING_CITY_VIEW"
    
    # 3. Other rules...
    pass

def extract_bed_type(room_name):
    if "King" in room_name:
        return "King"
    elif "Queen" in room_name:
        return "Queen"
    elif "Twin" in room_name:
        return "Twin"
    else:
        return "Unknown"

def extract_view_type(room_name):
    if "City View" in room_name or "City" in room_name:
        return "City View"
    elif "Sea View" in room_name or "Sea" in room_name:
        return "Sea View"
    elif "Garden View" in room_name or "Garden" in room_name:
        return "Garden View"
    else:
        return "Unknown"

def extract_room_class(room_name):
    if "Deluxe" in room_name:
        return "Deluxe"
    elif "Superior" in room_name:
        return "Superior"
    elif "Premium" in room_name:
        return "Premium"
    elif "Executive" in room_name:
        return "Executive"
    else:
        return "Standard"

3. Attribute Supplementing

Supplement Missing Attributes

def supplement_room_attributes(room, giata_room):
    # 1. If room has no bed type, supplement from GIATA
    if not room.bed_type:
        room.bed_type = giata_room.bed_type
    
    # 2. If room has no area, supplement from GIATA
    if not room.area:
        room.area = (giata_room.area_min + giata_room.area_max) / 2
    
    # 3. If room has no amenities, supplement from GIATA
    if not room.amenities:
        room.amenities = giata_room.amenities
    else:
        # Merge amenities
        room.amenities = list(set(room.amenities + giata_room.amenities))
    
    # 4. If room has no view, supplement from GIATA
    if not room.view:
        room.view = giata_room.view
    
    return room

4. Consistency Verification

Verify Room Mapping Accuracy

def validate_room_mapping(supplier_room, giata_room):
    warnings = []
    
    # 1. Verify bed type consistency
    supplier_bed = extract_bed_type(supplier_room.name)
    if supplier_bed != giata_room.bed_type:
        warnings.append(f"Bed type mismatch: {supplier_bed} vs {giata_room.bed_type}")
    
    # 2. Verify area consistency
    if supplier_room.area and giata_room.area_min:
        if supplier_room.area < giata_room.area_min * 0.9 or \
           supplier_room.area > giata_room.area_max * 1.1:
            warnings.append(f"Area mismatch: {supplier_room.area} vs {giata_room.area_min}-{giata_room.area_max}")
    
    # 3. Verify amenities consistency
    supplier_amenities = set(supplier_room.amenities)
    giata_amenities = set(giata_room.amenities)
    intersection = supplier_amenities & giata_amenities
    if len(intersection) / len(giata_amenities) < 0.7:
        warnings.append(f"Amenities mismatch: {len(intersection)}/{len(giata_amenities)}")
    
    # 4. Return verification result
    return {
        "is_valid": len(warnings) == 0,
        "warnings": warnings
    }

5. Manual Review (Optional)

Low Confidence Mappings Require Manual Review

def review_low_confidence_mappings():
    # 1. Query low confidence mappings (confidence < 0.85)
    low_confidence_mappings = database.query(
        "SELECT * FROM room_mappings WHERE confidence < 0.85 AND reviewed = false"
    )
    
    # 2. Manual review
    for mapping in low_confidence_mappings:
        # Display to reviewer
        display_mapping_for_review(mapping)
        
        # Wait for review result
        review_result = wait_for_review(mapping.id)
        
        if review_result.approved:
            # Approved, update mapping
            database.update(
                "UPDATE room_mappings SET reviewed = true, confidence = 1.0 WHERE id = ?",
                mapping.id
            )
        else:
            # Rejected, delete mapping
            database.delete(
                "DELETE FROM room_mappings WHERE id = ?",
                mapping.id
            )

Our Advantages

1. GIATA Authoritative Database

  • 500,000+ hotels
  • 1,000,000+ room types
  • Standardized room type names and attributes
  • Authoritative room mapping standard

2. AI Room Mapping

  • Word vector similarity
  • 85%+ accuracy
  • Automatic matching, no manual intervention needed

3. Rule Engine (Auxiliary)

  • Rule-based matching
  • Improves accuracy
  • Handles edge cases

4. Attribute Supplementing

  • Automatically supplements missing attributes
  • Unifies room type information
  • Improves user experience

5. Consistency Verification

  • Automatically verifies mapping accuracy
  • Identifies incorrect mappings
  • Low confidence mappings manually reviewed

Call to Action

Have You Seen the Same Room with 5 Different Names?

Problems:

  • 5 suppliers, 5 different room type names
  • Incomplete attributes (some suppliers missing bed type, area, amenities)
  • Semantic matching difficulties (Deluxe vs Superior vs Premium)
  • Price mismatch (false matching causes)
  • Users complain rooms don’t match photos

Our Solution:

  • GIATA authoritative database (standardized room type names and attributes)
  • AI room mapping (word vector similarity, 85%+ accuracy)
  • Attribute supplementing (automatically supplements missing attributes)
  • Consistency verification (automatically verifies mapping accuracy)
  • Manual review (low confidence mappings)

You only need:

import "github.com/hotelbyte-com/sdk-go/hotelbyte"

client := hotelbyte.NewClient("YOUR_API_KEY")

// Search hotels (automatically handles room mapping)
result, err := client.SearchHotels(&hotelbyte.SearchRequest{
    Destination: "London",
    CheckIn:     time.Date(2026, 2, 10, 0, 0, 0, 0, time.UTC),
    CheckOut:    time.Date(2026, 2, 12, 0, 0, 0, 0, time.UTC),
    Guests:      2,
})

// We automatically handle:
// - GIATA authoritative database query
// - AI room mapping
// - Attribute supplementing
// - Consistency verification

for _, hotel := range result.Hotels {
    for _, room := range hotel.Rooms {
        fmt.Printf("%s: $%.2f\n", room.Name, room.TotalPrice)
    }
}

No room type name chaos. No false matching. Only accurate room mapping.


Next Steps

Free Trial

  • 30-day free trial
  • No credit card required
  • Start testing immediately

Start Your 30-Day Free Trial

View Documentation

Contact Us

Have questions? Contact our engineers directly.

Contact Us


Series Navigation


Next up: Summary - Why Can We Solve These Problems?


Reading time: ~18 minutes Difficulty: Medium (requires understanding of room type mapping, GIATA, AI similarity)