Litintuit

An intelligent book recommendation system powered by DistilRoBERTa and vector embeddings.

Overview

Litintuit is a semantic book recommender that provides genre- and emotion-based book suggestions. Unlike traditional recommendation systems that rely on collaborative or content-based filtering, LitIntuit uses a pure LLM-driven approach to understand the emotional tone and themes of books, allowing for more intuitive and nuanced recommendations.

Key Features

• LLM-Powered Understanding – Uses DistilRoBERTa to analyze books and user queries at a semantic level.
• Free-Text Query Support – Users can describe their reading preferences naturally (e.g., "I want a hopeful and introspective coming-of-age novel").
• Emotion & Tone Matching – Beyond genre, LitIntuit recommends books based on their mood.
• Vector Search for Similarity – Utilizes embeddings to find books that closely align with user preferences.
• Simple CSV-Based Metadata Storage – No external APIs or databases—just efficient local data handling.

How It Works

1. User Input: The user enters a free-text query describing their ideal book.
2. Text Embedding & Vector Search: The system converts both the user query and book metadata into vector representations.
3. LLM-Powered Matching: Using DistilRoBERTa, the system identifies books with similar themes, tones, and genres.
4. Recommendation Output: The results are displayed in a ranked gallery grid, with up to 16 recommendations per query.

Technology Stack

• Natural Language Model: Hartmann's Emotion English DistilRoBERTa-base for semantic text analysis.
• Vector Embeddings: Used to represent book data and user queries in a similarity-based retrieval system.
• Data Storage: CSV files for storing book metadata, ensuring lightweight and flexible data handling.
• Frontend & Backend: Built as a web application for a user-friendly experience using Python and Gradio.

Getting Started

To run Litintuit locally:

• Clone this repo.
• Run python gradio-dashboard.py in your terminal.
• The server will take some time to load due to data chunking. Wait for approximately 3 minutes.
• After some time, the server will display the local url where it is hosted. Open the web app in your browser and start discovering books! 📚