AI Book Advisor

Coming SoonGitHub
AI Book Advisor

Welcome to AI-Based Hybrid Book Recommendation System

An AI-powered book recommendation system that combines Cosine Similarity, Neural Collaborative Filtering (NCF), and CNN-based text modeling to generate diverse and meaningful book recommendations.

The project processes a dataset of 110K+ books, applies multiple machine learning techniques, and combines their outputs into a single hybrid recommendation pipeline.

Introduction

Finding relevant books from thousands of available titles can be challenging. Traditional recommendation systems often rely on a single source of information, such as book content or user ratings.

This project explores a hybrid approach, combining content similarity, user–book relationships, and textual patterns to create a more comprehensive recommendation system.

The complete workflow was divided into multiple processing stages and notebooks because of the computational and memory requirements of the models.

Model Flow

                Book Data
                    ↓
              Preprocessing
                    ↓
┌─────────────┬─────────────┬─────────────┐
│   Cosine    │     NCF     │     CNN     │
│ Similarity  │    Model    │ Text Model  │
└─────────────┴─────────────┴─────────────┘
                    ↓
              Hybrid Ranking
                    ↓
           Book Recommendations

Key Features

  • Hybrid Recommendations: Combines three different recommendation approaches into one system.
  • Content-Based Filtering: Uses TF-IDF and cosine similarity to find books with similar titles, descriptions, and genres.
  • Neural Collaborative Filtering: Uses user and book embeddings to learn relationships between users and books.
  • CNN Text Model: Uses book descriptions to learn textual patterns for rating prediction.
  • Large-Scale Processing: Processes and filters a dataset containing 110K+ books, resulting in approximately 56K books for the final pipeline.
  • Interactive Interface: Built a Streamlit interface where users can select a book and explore recommendations with book details.

Technologies Used

  • Machine Learning: Scikit-learn, TensorFlow / Keras
  • Recommendation: TF-IDF, Cosine Similarity, NCF
  • Deep Learning: Embeddings, CNN, Neural Networks
  • Data Processing: Pandas, NumPy
  • Deployment / UI: Streamlit
  • Compute: GPU & TPU
  • Experimentation: Kaggle Notebooks

How It Works

The system first preprocesses the book dataset and prepares the information required by each model.

The Cosine Similarity model identifies books with similar textual content, while the NCF and CNN models generate additional prediction signals.

The recommendations from all three models are then merged, duplicates are removed, and the results are ranked to produce the final recommendation list.

The computationally intensive stages were separated into individual notebooks for easier processing and model management.

┌──────────────────┐ ┌──────────────────┐ ┌──────────────────┐
│ COSINE           │ │ NCF              │ │ CNN              │
│ SIMILARITY       │ │                  │ │                  │
│                  │ │ User + Book      │ │ Book Description │
│ TF-IDF           │ │ Embeddings       │ │ Text Model       │
│                  │ │                  │ │                  │
│ Content Signal   │ │ User Signal      │ │ Text Signal      │
└──────────────────┘ └──────────────────┘ └──────────────────┘
                         ↓
                   HYBRID MODEL
                         ↓
                  RECOMMENDATIONS

Explore the Project

  • Project Walkthrough: Watch the complete video explaining the architecture, models, processing pipeline, and Streamlit implementation.