Brian Norlander / projects /

Song Sentiment Predictor

Published on May 01, 2018.

This was a partner project for my final project for an artificial intelligence class (CSCI 4511W) that I took during the spring of 2018 at the University of Minnesota.

The purpose of this project was to predict the sentiment of a song. There were 7 categories: happy, anger, funny, hurt, calm, romantic, and inspirational. Our program used the Naive Bayes and Support Vector Model algorithms to create a classifier to predict the sentiment of songs based only on its lyrics. The ground truth was based on 400+ songs scraped from lyrica. Once the classifiers were created we built an interactive web page where a user could enter a song name and the predicted sentiment would display.

View on GitHub