📚 NextLeaf — AI Semantic Book Recommendation System

Discover your next favorite book using Semantic Search, Vector Embeddings, and Emotion-Aware Ranking.

---

📖 Overview

NextLeaf is an AI-powered semantic book recommendation system that recommends books based on meaning rather than keywords.

The application uses BAAI/bge-small-en-v1.5 embeddings with ChromaDB to retrieve semantically similar books and then improves the results using emotion-aware ranking generated from book descriptions.

---

✨ Features

• 🔎 Semantic search using vector embeddings
• 📚 Content-based recommendations
• ❤️ Emotion-aware ranking
• 🗂️ Genre filtering
• ⭐ Rating filtering
• ⚡ ChromaDB vector database
• 🤖 Hugging Face Transformers
• 🎨 Interactive Gradio interface
• 📱 Responsive UI

---

🏗️ System Architecture

User Query
     │
     ▼
Gradio Interface
     │
     ▼
BGE Embedding Model
     │
     ▼
ChromaDB Vector Search
     │
     ▼
Candidate Books
     │
     ▼
Emotion-aware Re-ranking
     │
     ▼
Recommended Books

---

🧠 Recommendation Pipeline

1. User enters a natural language query.
2. The query is converted into an embedding.
3. ChromaDB retrieves semantically similar books.
4. Apply optional filters (Category, Mood, Rating).
5. Emotion scores re-rank the retrieved books.
6. Display the best recommendations.

---

🛠️ Tech Stack

Category	Technology
Language	Python 3.12
UI	Gradio
Framework	LangChain
Vector Database	ChromaDB
Embeddings	BAAI/bge-small-en-v1.5
NLP	Hugging Face Transformers
Data Processing	Pandas

---

📂 Project Structure

NextLeaf/
├── app.py
├── create_chroma_db.py
├── README.md
├── requirements.txt
├── LICENSE
├── .gitignore
├── data/
│   ├── chroma_db/
│   ├── books_cleaned.csv
│   ├── books_with_categories.csv
│   ├── books_with_emotions.csv
│   └── tagged_description.txt
└── notebooks/
    ├── data-exploration.ipynb
    ├── sentiment-analysis.ipynb
    ├── text-classification.ipynb
    └── vector-search.ipynb

---

⚙️ Installation

git clone https://github.com/<your-username>/NextLeaf.git
cd NextLeaf

python -m venv .venv
source .venv/bin/activate    # macOS/Linux

# Windows
# .venv\Scripts\activate

pip install -r requirements.txt
python app.py

---

📚 Dataset

This project uses the 7K Books with Metadata dataset from Kaggle.

The dataset includes:

• Book title
• Author
• Description
• ISBN
• Categories
• Average rating
• Thumbnail

Book descriptions are enriched using the j-hartmann/emotion-english-distilroberta-base model.

---

🔍 Semantic Search

Instead of matching exact words, NextLeaf retrieves books with similar meaning using dense vector embeddings.

Example:

"A magical coming-of-age adventure with friendship and danger"

---

❤️ Emotion-aware Ranking

Each book description is classified into:

• Joy
• Fear
• Sadness
• Anger
• Surprise
• Disgust
• Neutral

When a mood is selected, semantic similarity and emotion scores are combined to improve recommendation quality.

---

🚀 Future Improvements

• User authentication
• Reading history
• Collaborative recommendations
• Hybrid BM25 + Vector Search
• LLM-generated summaries
• Docker support
• Cloud deployment

---

🙏 Acknowledgements

• LangChain
• Hugging Face
• ChromaDB
• Gradio
• Kaggle

---

👨‍💻 Author

Daksh Kadyan

B.Tech Computer Science & Engineering
Delhi Technological University (DTU)

---

⭐ Support

If you found this project useful, consider giving the repository a ⭐ on GitHub.