One-Day Tour Planning Helper

Hola! This section provides a comprehensive overview of the architecture and components used to build this application.

Project Architecture

System Overview

The One-Day Tour Planning helper is built using a modular and scalable architecture, incorporating multiple agents for specific tasks, memory management, and a user-friendly frontend interface. Below is a detailed breakdown:

---

Key Components

1. Frontend:

   • Framework: Streamlit.
   • Features:
     • A user login system for personalized access.
     • An interactive chat interface for planning discussions.
     • A display area for itineraries and real-time updates.

2. Backend:

   • Framework: FastAPI.
   • Microservices:
     • User Interaction Agent: This agent manages user inputs and guides the conversation, ensuring a smooth interaction.
     • Itinerary Generation Agent: Responsible for crafting personalized itineraries based on user preferences.
     • Optimization Agent: Provides current weather information and adjusts plans as needed.
     • Weather Agent: Provides current weather information and adjusts plans as needed.
     • Memory Agent: Utilizes Neo4j to store and recall user preferences, enhancing personalization.

3. Memory Management:

   • Database: Neo4j (GraphDB).
   • Functionality:
     • Stores user preferences as triplets (Entity-Relationship-Entity).
     • Adapts dynamically to user inputs for seamless personalization.

4. Tools and Libraries:

   • Language Models: Transformers/Ollama/vLLM for LLM endpoints.
   • Database: Neo4j for memory graph.
   • API Framework: FastAPI for backend microservices.
   • Frontend: Streamlit for chat and display.

---

Workflow

1. User Interaction:

   • Users input details such as city, budget, interests, and timings.
   • The User Interaction Agent collects this information and guides the planning process.

2. Memory Integration:

   • User preferences are stored in Neo4j, allowing the system to recall them during future interactions.
   • The system adapts dynamically to ensure a personalized experience.

3. Itinerary Creation:

   • The Itinerary Generation Agent suggests tailored plans based on the collected inputs.
   • Adjusments are made t fit within the specfied budget and time constraints.

4. Weather and News Updates:

   • The Weather Agent provides real-time weather forecasts relevant to the user's destination.
   • Alerts about news or events related to the location are also fetched.

5. Dynamic Updates:

   • Users can modify their preferences during the conversation.
   • The system instantly re-optimizes the itinerary to reflect these changes.

---

Architecture of this project

graph TD
    A[Frontend: Streamlit] --> B[Backend: FastAPI]
    B --> C[User Interaction Agent]
    B --> D[Itinerary Generation Agent]
    B --> E[Optimization Agent]
    B --> F[Weather Agent]
    B --> G[Memory Agent: Neo4j]
    G --> H[GraphDB: User Preferences]
    C -->|Inputs| G
    E -->|Optimized Paths| G
    D -->|Itineraries| A

Loading