Deep Recommendation Systems on Amazon Reviews

This repository contains research implementations of deep recommendation systems utilizing Amazon reviews. It features multiple deep learning models that leverage review texts for rating prediction.

---

📌 Project Overview

This project explores various deep learning approaches to recommendation systems by integrating textual features with collaborative filtering techniques. The implemented models include:

🔹 DeepCoNN

A dual Convolutional Neural Network (CNN) model that processes separate user and item reviews, combining their representations for rating prediction.

🔹 HSACN

A hierarchical review embedding model that extracts n-gram features from reviews via convolutional operations while analyzing review statistics.

🔹 NRCMA

A model incorporating cross-attention mechanisms and factorization machines to capture complex interactions between textual review features and identity embeddings.

---

📂 Dataset Details

The models are trained and evaluated using a subset of the publicly available Amazon Reviews 2023 dataset from the McAuley-Lab. Specifically, the "raw_review_Appliances" subset is used for this project.

---

📁 Repository Structure

├── deep-rec-sys-amazon-reviews/
│   ├── pyproject.toml         # Project configuration and dependencies
│   ├── uv.lock                # Environment lock file
│   ├── assets/
│   │   └── images/            # Visual assets, plots, and figures
│   ├── eda/                   # Exploratory Data Analysis (EDA) notebooks
│   │   ├── Amazon_user_reviews_capstone.ipynb
│   │   ├── item_reviews.ipynb
│   │   └── recsys_preprocessing.ipynb
│   ├── modeling/              # Model implementations and training scripts
│   │   ├── DeepCoNN.ipynb
│   │   ├── DeepCoNN.py
│   │   ├── HSACN.ipynb
│   │   ├── HSACN.py
│   │   ├── NRCMA.ipynb
│   │   └── NRCMA.py

Each component is structured to separate data analysis from model development, ensuring clarity and ease of extension.

---

🚀 Installation and Setup

🔹 Clone the Repository

git clone https://github.com/your-username/deep-rec-sys-amazon-reviews.git
cd deep-rec-sys-amazon-reviews

🔹 Create a Virtual Environment

It is recommended to use Python 3.12 or later in a virtual environment:

python -m venv env
source env/bin/activate  # On Windows: env\Scripts\activate

🔹 Install Dependencies

Install the required packages from pyproject.toml or requirements.txt:

pip install -r requirements.txt

Alternatively, manually install key dependencies:

pip install gensim>=4.3.3 matplotlib>=3.10.0 nltk>=3.9.1 numpy<2.0.0 \
            pandas>=2.2.3 scikit-learn>=1.6.1 surprise>=0.1 torch>=2.6.0 \
            torchinfo>=1.8.0 torchvision>=0.21.0 tqdm>=4.67.1 wandb>=0.19.6

Ensure your Python version meets the required (>=3.12) specification.

---

⚡ Running the Code

🔹 Exploratory Data Analysis (EDA)

Navigate to the eda/ directory and open notebooks using Jupyter:

jupyter notebook eda/Amazon_user_reviews_capstone.ipynb

These notebooks cover preprocessing, data visualization, and statistical analysis of Amazon user reviews.

🔹 Model Training

To train a model, navigate to the modeling/ directory and run the corresponding Python script:

python modeling/DeepCoNN.py  # Example for DeepCoNN

Each script handles data loading, preprocessing, training, evaluation, and model saving. Refer to notebooks for additional details on hyperparameters and configurations.

---

📚 Additional Resources

For a deeper understanding of the algorithms, evaluation metrics, and experimental setup, visit the project’s website:
🔗 Project Website

---

🤝 Contributing

Contributions are welcome! If you’d like to improve the code, models, or documentation, feel free to open an issue or submit a pull request.

---

📜 License

This project is open source. Refer to the LICENSE file for details on usage and distribution rights.

---

🔹 Happy Coding! 🚀