Bookish 📚

A smart book recommender

An AI-powered book recommendation system leveraging Semantic Search, Zero-Shot Text Classification and Sentiment Analysis.

check it out here

Features ✨

• Get personalised book recommendations based on your query.
• Filter books based on the category of Fiction/Nonfiction.
• Filter books based on the Genre of the book.
• Get the description along with a link to google search results.

Structure

Bookish/
├── .devcontainer/          # for GitHub spaces
├── .streamlit/             # for custom theme
├── chroma_db/              # vector database 
├── datasets/               # Book datasets (csv)
├── image/                  # images for readme
├── notebooks/              # steps to build the project
├── utils/                  # functions doing the work
├── stream_interface.py     # the main file
├── requirements.txt        # requirements
├── README.md               # Simple docs
└── LICENSE                 # you can contribute

---

Mechanics ⚙️

Semantic Search

• Uses an Open Source sentence-transformer model for creating the vector database from the book dataset.
• Performs similarity search between your query and the books vector database.

Zero-Shot Text Classification

• Uses an Open Source zero-shot text classification model to classify books into Fiction/Nonfiction Category.
• This way helps in filling-in the missing data from the dataset.

Sentiment Analysis

• Uses a sentiment analysis model to classify books based on the emotional tone of the description.
• Calculates a score from 0-1 for Ekman's 6 basic emotions from the book description.

Contribution

This project will remain open for contribution, please feel free to help.

Acknowledgments

Thanks to Arjav for helping out with the CSS part and Dr. Jodie Burchell for being such a great instructor.

This project is based on an OpenSource book dataset from Huggingface.

---

made with 🫶🏻