39 lines (32 loc) · 1.8 KB

TravelBot Documentation

Core System Documentation

🧠 Travel Recommendation RAG System

The heart of TravelBot - A sophisticated Retrieval-Augmented Generation (RAG) system that combines:

TravelQueryAnalyzer for natural language understanding
VectorSearchService for semantic similarity search
SearchResultRanker for multi-criteria result ranking
RAGContextBuilder for context aggregation
TravelPreferenceTracker for user personalization

🔍 Vector Search & AI System

PostgreSQL pgvector implementation with AWS Bedrock integration:

pgvector extension with HNSW indexes for fast similarity search
AWS Bedrock Titan V2 embeddings (1024 dimensions)
Semantic search across destinations, resorts, and amenities
AI-powered data seeding system
Async embedding generation with Symfony Messenger

Technology Stack

Component	Technology
Backend	Symfony 7.3, PHP 8.1+ (8.3 in Docker), Doctrine ORM
AI	Claude AI via AWS Bedrock, Titan V2 Embeddings
Database	PostgreSQL with pgvector extension (v15 local, Neon cloud in production)
Frontend	Twig, Hotwire Turbo, Tailwind CSS 4
Infrastructure	AWS ECS Fargate, CloudWatch, Neon PostgreSQL

API Endpoints

The system provides several REST API endpoints documented in the RAG system documentation:

POST /chat/query-analysis - Analyze travel queries
GET /chat/personalized-recommendations - Get personalized recommendations
POST /chat/track-interaction - Track user interactions
GET /chat/conversation-context/{id} - Get conversation context
GET /chat/smart-suggestions/{id} - Get contextual suggestions
GET /chat/stream/{id} - Server-sent events for streaming responses