GenAI Route Recommendation for Travel

This repository contains the source code and documentation for a GenAI-powered web application designed to deliver personal> Tip

The URLs are defined via Helm values.yaml – adjust the ingress.host field of each chart if you deploy to a different domain.

Documentation

Additional detailed documentation, architectural diagrams, and API specifications can be found in the project's Confluence page.ravel itineraries using conversational AI to provide personalized attraction recommendation.

Overview

Travel planning can be complex and tedious. This project addresses the issue by creating highly personalized itineraries, optimizing according to multiple user preferences, and offering real-time, reliable recommendations via AI.

Intended Users

• Independent Travelers seeking personalized travel routes.
• Industry Partners (such as Ticket Booking Platforms) aiming to enhance their services with personalized route planning.

Main Functionalities

1. Conversational Trip Builder

Interact naturally with an AI assistant to generate customized travel routes. Example:

• User input: "I have 5 days in Munich, architect fan and want to explore authentic local restaurants."
• Output: Personalized travel recommendation with explanation.

2. Preference-Aware Optimization

Routes are optimized based on user-defined preferences such as budget, scenic value, and accessibility. Preferences are persistently stored for continuous improvement over multiple sessions.

3. Reliable Recommendations

The system employs Retrieval-Augmented Generation (RAG) to recommend real-world attractions backed by current and verified information from reliable sources.

System Architecture

• Client Side: Built with React.js, interacts with users via intuitive conversational UI.
• Server Side: Spring Boot for core business logic.
• AI Components: GenAI model (LLMs with RAG integration) for conversational reasoning and recommendations.
• Persistence Layer: PostgreSQL database for storing user preferences and session data.

Installation and Setup

Prerequisites

• Java 21 (for development setup)
• Node.js 18 or higher (for development setup)
• PostgreSQL (for development setup)
• Docker and Docker Compose (for Docker deployment - recommended)
• Python 3.11+ with Poetry (for GenAI service development)

Steps

Docker Setup (recommended)

1. Clone Repository

   git clone git@github.com:AET-DevOps25/team-drop-database.git
   cd team-drop-database

2. Configure Environment Variables

   Set up the necessary environment files for each service:

   # Copy example environment files and configure them
   cp genai/.env.example genai/.env
   # Edit genai/.env file with your OpenAI API key and other configurations
   
   # Other services use environment variables defined in docker-compose.yml

3. One-Click Docker Deployment

   For automatic deployment of all services:

   # Option 1: Use the provided script (recommended)
   ./start-docker-all.sh
   
   # Option 2: Use docker-compose directly
   docker-compose up --build

   This will automatically:

   • Build and start all microservices (AuthService, AttractionService, UserService)
   • Start the GenAI service
   • Start the React frontend
   • Set up PostgreSQL database with initial schema
   • Configure all service networking and dependencies

Development Setup

1. Clone Repository

   git clone git@github.com:AET-DevOps25/team-drop-database.git

2. Server Side Setup

   cd server/AttractionService
   ./gradlew build

   cd server/AuthService
   ./gradlew build

   cd server/UserService
   ./gradlew build

   before running the microservices, make sure to set up the PostgreSQL database and update the application.properties files with your database credentials. you should also specify the profile to be "dev"

3. GenAI Service Setup

   cd genai

   Create and activate a virtual environment:

   # Using venv (recommended)
   python -m venv venv
   source venv/bin/activate  # On macOS/Linux
   
   # Or using conda
   conda create -n travelbuddy python=3.11
   conda activate travelbuddy

   Install dependencies:

   # Install poetry if not already installed
   brew install poetry  # On macOS
   
   # Install project dependencies
   poetry install
   
   # Download required language model
   python -m spacy download en_core_web_sm

   Set up environment variables:

   cp .env.example .env
   # Edit .env file with your API keys and configuration

4. Frontend Setup

   cd client/travel-buddy
   npm install
   npm start

Testing

The project includes comprehensive testing for all components:

Running Tests

# Frontend tests only
cd client/travel-buddy
npm run test:ci

# Or use the convenience script
./test-frontend.sh

# Backend Java tests
cd server/AttractionService && ./gradlew test
cd server/AuthService && ./gradlew test  
cd server/UserService && ./gradlew test

# Python/GenAI tests
cd genai && pytest -v

CI/CD Testing

All tests run automatically in GitHub Actions on every push and pull request. The pipeline includes:

• Frontend React/TypeScript tests with Jest
• Backend Java unit tests with Gradle
• Python tests with pytest

Usage

• Access the application via http://localhost:3000 or under https://travel-buddy.student.k8s.aet.cit.tum.de/
• Begin by providing preferences or directly interacting with the conversational AI.

Student Responsibilities

Student Name	Responsibility
Shuaiwei Yu	1. AuthService Implementation 2. Client Side Authentication & Conversation Page Implementation 3. Helm and K8S deployment
Haochuan Huai	1. GenAI Service Implementation 2. AWS Deployment 3. Backend Attraction Service API Implementation 4. Frontend Attraction Detail Page Implementation
Zhiyuan Ni	1. Attraction Service Implementation 2. Grafana and Prometheus Implementation 3. Helm and K8S Deployment and Debugging 4. Frontend List View Implementation

Monitoring & Service Endpoints

Once the Helm charts are deployed on the TUM student cluster, the following public URLs are available:

Component	URL
Frontend	https://travel-buddy.student.k8s.aet.cit.tum.de
Grafana (dashboards, alerting)	https://grafana.travel-buddy.student.k8s.aet.cit.tum.de
Prometheus (raw metrics & targets)	https://prometheus.travel-buddy.student.k8s.aet.cit.tum.de
Auth-Service API	https://auth-service.travel-buddy.student.k8s.aet.cit.tum.de/auth-service/api/v1
User-Service API	https://user-service.travel-buddy.student.k8s.aet.cit.tum.de/user-service/api/v1
Attraction-Service API	https://attr-service.travel-buddy.student.k8s.aet.cit.tum.de/attraction-service/api/v1
LLM-Service API	https://llm-service.travel-buddy.student.k8s.aet.cit.tum.de/api/v1

Tip
The URLs are defined via Helm values.yaml – adjust the ingress.host field of each chart if you deploy to a different domain.

Documentation

Additional detailed documentation, architectural diagrams, and API specifications can be found in the project’s Confluence page.