Demo of the work

Recommendation_Engine_Demo.mp4

Project Title: Goodreads Book Recommendation System

Overview

This project presents an advanced book recommendation system utilizing a hybrid approach that amalgamates both item-based and user-based filtering techniques. Leveraging data scraped from Goodreads, the system applies sophisticated data processing and machine learning methodologies to deliver personalized book recommendations to users.

Development Process

Process	Description	Potential Improvements
Web Scraping from Goodreads	Implemented asynchronous web scraping techniques to efficiently gather comprehensive book data from Goodreads.
Data Cleaning and Preprocessing	Employed robust Natural Language Processing (NLP) techniques and rigorous data cleaning methodologies to ensure data quality.
Content-Based Filtering and Synthetic Data Generation	Integrated content-based filtering techniques, augmented with synthetic data generation to handle missing values.
Feature Engineering	Utilized advanced feature engineering strategies to extract meaningful insights from the dataset.
User and Item Profile Creation	Engineered user and item profiles based on book genres, employing dimensionality reduction techniques for enhanced performance.
Flask Application Development	Developed a robust Flask application featuring a secure login system and dynamic book recommendations.
Data Storage and Retrieval	Utilized PostgreSQL database for efficient data storage and retrieval, employing a combination of API calls and SQL queries.
Model Serialization and Deployment	Serialized trained models for streamlined deployment using Gunicorn.
Dockerization	Containerized the application for optimized deployment and scalability.
CI/CD	Established a robust CI/CD pipeline using Docker, GCP, and GitHub Actions for seamless integration and deployment.

Things That Can Be Improved

1. Middleware and Caching: Implement middleware for input validation and utilize Redis for caching database calls to improve performance and efficiency.Adding Logging and testing too would be great for the implementation.

2. Advanced Techniques: Explore advanced recommendation techniques such as Neural Collaborative Filtering, Alternating Least Squares (ALS), and Decision Trees (content-based classifier) to enhance recommendation accuracy and diversity.

3. Implementing Micro-service Design: Adding Microservices is a better and scalable approach which will use the full capabilities of Cloud, Adding ways to implement database transactions,event Sourcing db,Chuncking big db(AWS_RDS/Google_cloudsql or AWS_S3 ).

Getting Started

To begin using this project:

1. Clone the repository.

2. Install dependencies using Poetry:

   poetry install

3. Run the application using Gunicorn:

   poetry run gunicorn -w 4 -b 0.0.0.0:8000 main:app

4. Login using Userid and Password

   The user id and password are given in JSON FILE

License

This project is licensed under the MIT License.

Resources

• GCP documentaion
• Flask
• Postgresql
• BART documentaion
• Docker documentation
• Doc2vec
• User Profiling
• CI-CD pipelines using gcp
• Other approaches to build recommendations