Date of Award

6-2023

Document Type

Thesis - SCU Access Only

Publisher

Santa Clara : Santa Clara University, 2023.

Department

Computer Science and Engineering

First Advisor

Yi Fang

Abstract

Many popular media recommendation algorithms suffer from over recommending the most popular results. We wanted to create a music recommendation system that relied completely on song data and no demographic data to provide less biased song results. We created a Python and Flask web application that takes data from Spotify about songs and targets similar songs using the K-Nearest Neighbors algorithm. Our user interface allows users to login to Spotify, select songs as seeds for our algorithm, and then examine data about the recommended results. Our algorithm provided recommendations that were rated on average half as popular as Spotify’s algorithm, and provided almost twice as much coverage of long-tail items. The algorithm shows promising results, and with further testing, computing power, and time, it could be improved even more.

Download

SCU Access Only

To access this paper, please log into or create an account in Scholar Commons using your scu.edu email address.

Share

COinS