Skip to content

README.md

Latest commit

 

History

History
64 lines (49 loc) · 4.12 KB

README.md

File metadata and controls

64 lines (49 loc) · 4.12 KB
Raw

Entirary - Your Personalized Travel Discovery Platform

License: MIT Maintenance Deployed on Netlify

✨ Discover Your Perfect Getaway with Precision ✨

Entirary isn't just another travel recommendation engine. We leverage the power of Retrieval-Augmented Generation (RAG) to understand your unique travel preferences and match you with destinations that truly resonate. Forget generic lists – Entirary dives deep to find hidden gems and popular spots that align with your individual tastes.

🚀 Key Features

  • Personalized Recommendations: Our RAG-based system analyzes your preferences with remarkable accuracy, going beyond basic filters.
  • Precise Similarity Matching: We use vector embeddings to understand the semantic similarity between your preferences and destination characteristics.
  • Rich Destination Information: Get comprehensive details about recommended locations, helping you make informed decisions.
  • Intuitive User Interface: Easily input your preferences and explore tailored suggestions powered by React.
  • Seamless Exploration: Discover new possibilities you might never have considered.

⚙️ How It Works

Entirary employs a sophisticated process to deliver your ideal travel recommendations:

  1. Preference Input: You tell us about your ideal trip – your interests (e.g., hiking, museums, food), travel style (e.g., budget-friendly, luxurious, adventurous), desired atmosphere (e.g., relaxing, vibrant, historical), and more.
  2. Embedding Generation: Your preferences are converted into a numerical vector embedding using Google Gemini via the @google/generative-ai library. This embedding captures the semantic meaning of your input.
  3. Vector Database Retrieval: This preference embedding is then used to query our PostgreSQL database (with the vector extension) to find destination embeddings that are semantically similar. We use the cosine distance (<->) to measure this similarity.
  4. Recommendation Generation (RAG): The retrieved destination information, along with your original preferences, is used by a generative model (likely Google Gemini) to create personalized recommendations and descriptions, highlighting why these destinations are a good fit for you.
  5. Tailored Suggestions: You receive a curated list of destinations, ordered by their similarity to your preferences, with detailed explanations.

🚀 Try It Out!

Visit the deployed website to experience personalized travel recommendations:

https://entirary.netlify.app/

🛠️ Technologies Used

  • Frontend: React.js
    • UI Components: @radix-ui/react-*, lucide-react
    • Styling: tailwindcss, tailwindcss-animate
    • Forms: react-hook-form, formik, yup, zod
    • Data Fetching: axios
    • Routing: react-router-dom
    • Animations: motion, popmotion
    • Other: date-fns, react-icons, react-spinners, react-toastify, clsx, rc-slider
  • Backend:
    • Framework: express
    • LLM & Embeddings: @google/generative-ai (using Google Gemini for both text generation and embeddings with embedding-001 model)
    • Database: PostgreSQL with the vector extension
    • Database Interaction: pg-promise
    • Authentication: bcrypt, jsonwebtoken
    • Middleware: cors
    • Environment Variables: dotenv
    • CSV Parsing: csv-parser
    • Development: nodemon
  • Vector Search: Utilizes the vector extension in PostgreSQL with cosine distance (<->) for efficient similarity search.

🗺️ Future Enhancements

  • Integration with booking platforms.
  • User accounts to save past trips.
  • More granular preference options.
  • Visualizations of destination similarities.
  • Community features for sharing travel experiences.
  • Support for multiple languages.