SQL and AI-Driven Personalized Content Recommendation Systems for Small Businesses and Streaming Services

Overview

This repository contains Egemen Özen's master's thesis, "Using SQL and AI for Personalized Content Recommendations in Streaming Services". The thesis delves into the development and adaptation of personalized recommendation systems leveraging SQL databases and AI algorithms. The aim is to bridge the gap between large-scale systems used by giants like Netflix and Amazon Prime, and their implementation for small-scale e-commerce businesses, using real-world case studies.

Abstract

The thesis provides an extensive examination of how personalized recommendation systems can be optimized for different scales, emphasizing data management and performance aspects. Key contributions include a comparative analysis of recommendation algorithms, an in-depth study of the role of SQL databases in data handling, and a focus on ethical considerations such as algorithmic bias, transparency, and data privacy. Real-world scenarios involving streaming services and small businesses illustrate how scalable recommendation engines can transform customer engagement and satisfaction.

Table of Contents

• Overview
• Abstract
• Contents
• Objectives
• Methodology
• Key Findings
• Technologies and Tools Used
• Key Challenges and Ethical Considerations
• How to Access and Cite
• Author
• Acknowledgements

Contents

• Thesis Document: Dissertation_EgemenOzen_Q1057283.pdf
• Chapters Include:
  • Literature Review of Recommendation Systems
  • AI Techniques for Content Personalization
  • Comparative Analysis of SQL vs. NoSQL Databases
  • Case Studies: Netflix, Amazon Prime, and Small Business Implementation
  • Ethical Considerations in AI-Driven Recommendations
  • Methodologies and Results from Case Studies

Objectives

1. Adaptation of Recommendation Algorithms: Explore methodologies to scale down recommendation systems for small businesses.
2. SQL's Role in Data Management: Analyze SQL database performance in handling customer and content data for recommendations.
3. Ethical AI: Address ethical issues such as data privacy, bias, and transparency in AI-driven recommendation systems.
4. Practical Case Studies: Draw lessons from industry leaders and translate them to small-scale e-commerce use cases.

Methodology

The research adopts a mixed-methods approach:

• Literature Review: A comprehensive overview of the evolution of recommendation algorithms.
• Primary Data Analysis: Real-world case study analysis using data from a small bakery business and comparisons with Netflix and Amazon Prime.
• SQL and AI Techniques: Demonstrates practical SQL-based data management and AI-based recommendations.
• Ethical Considerations: A dedicated analysis on data privacy, algorithmic fairness, and transparency.

Key Findings

• Small-scale businesses can leverage hybrid recommendation systems to improve customer engagement and sales.
• SQL databases provide robust capabilities for structured data management, but a hybrid approach with NoSQL offers superior flexibility for dynamic data.
• Ethical considerations such as transparency, fairness, and user privacy must be integral to the design of recommendation systems.

Technologies and Tools Used

• SQL Databases (MySQL, PostgreSQL)
• Python (Pandas, Scikit-learn, TensorFlow)
• Machine Learning Models: Collaborative filtering, content-based filtering, hybrid methods, deep learning approaches.
• Data Visualization: Matplotlib, Seaborn
• Real-World Case Studies: Netflix, Amazon Prime, and a local bakery dataset.

Key Challenges and Ethical Considerations

• Algorithmic Bias: Ensuring fair recommendations without perpetuating biases inherent in training data.
• Data Privacy: Complying with GDPR and other regulations.
• Model Interpretability: Addressing the "black box" nature of AI models with tools such as LIME and SHAP.
• Scalability: Adapting models designed for large-scale data to smaller datasets.

How to Access and Cite

To explore the thesis, download the PDF file included in this repository. If you wish to reference this work, please use the following citation format:

Author

Egemen Özen
Email: [[email protected]]
LinkedIn: [www.linkedin.com/in/egemen-özen]
GitHub: [https://github.com/egemenozen1]