Behind the scenes of nendeni.com's AI engine: how machine learning matches travelers to community tourism experiences based on budget, interests, season, and community benefit.
The Problem AI Solves
Mozambique has 51+ verified community tourism listings spread across 11 provinces — from Gorongosa safaris to Tofo diving, from Makonde woodcarving workshops to Ilha de Moçambique homestays. But most international travelers don't know which experience fits their budget, interests, or travel season.
The result? They default to Booking.com, where community-owned operators are invisible.
How AI Itinerary Matching Works
Nendeni.com's AI Demand Engine uses TF-IDF (Term Frequency-Inverse Document Frequency) combined with cosine similarity to match traveler profiles to listings. Here's what happens when a traveler says *"I want to see elephants in July, budget $3,000"**:
Step 1: Profile Parsing
The engine extracts structured data from natural language:
- Budget: $3,000 USD
- Travel month: July (month 7)
- Interests: wildlife, conservation, community
- Community priority: 0.9 (high — cares deeply about community benefit)
Step 2: Document Building
Every listing becomes a searchable document combining its name, province, category, description, and tags. The engine builds a TF-IDF matrix across all listings — essentially creating a mathematical representation of what each experience offers.
Step 3: Seasonal Validation
The engine checks whether July is a valid season for each listing. Gorongosa? Peak dry season — optimal game driving. Tofo whale sharks? Wrong season (October is peak). This filter eliminates unsuitable matches immediately.
Step 4: Composite Scoring
Each remaining listing receives a composite score across five dimensions:
| Factor | Weight | What It Measures | |--------|--------|-----------------| | Text relevance | 35% | How well listing matches traveler interests | | Seasonal availability | 20% | Is this the right time to visit? | | Budget fit | 15% | Can the traveler afford this experience? | | Community benefit | 20% | Does this listing align with community priority? | | Quality rating | 10% | What do other travelers say? |
Community-owned operators with 100% community benefit receive a natural boost — the engine is designed to favor ethical tourism.
Step 5: Itinerary Generation
The top matches are routed using nearest-neighbor geographic routing (haversine distance) to create a logical day-by-day itinerary. Google Maps deep links are generated for each leg of the journey.
Why This Matters for Mozambique
This is the core of what ANDITUR (Mozambique's National Agency for Tourism Development and Investment) needs: a system that doesn't just list tourism offerings but actively matches them to the right travelers.
Traditional tourism platforms are passive — they wait for travelers to search. AI-driven matching is proactive — it understands intent and recommends accordingly. For community tourism, this is revolutionary because rural operators can't compete with international hotel chains on SEO alone. The AI levels the playing field by surfacing community experiences that match traveler values.
Demand Intelligence for National Planning
Beyond individual matching, the engine aggregates anonymized data across all traveler profiles to generate demand insights: which provinces are most requested, which interests dominate, budget distribution, peak travel months. This intelligence feeds directly into ANDITUR's national tourism strategy.
When 70% of analyzed profiles show "community priority" above 0.7, that's a signal: Mozambique's tourism future isn't mass tourism — it's ethical, community-centered tourism. The data proves it.
The AI Demand Engine is part of nendeni.com's Technology page. Explore the platform or try the AI Concierge to see it in action.