Ward, Matthew Oliver
Currently, there are several methods for recommending music to an individual. These vary from word-of-mouth to recommendation databases. This project attempts to visualize the lyrical relationships found in collections of songs. We discover these relationships by examining the structural and contextual components of a song's lyrics, and using those components in a comparison algorithm. We gathered a set of 5605 lyrics and associated data in order to establish a network of relationships to navigate and explore. The visualizations allow users to graphically explore the collection to discover trends, interesting data, or to find new songs to listen to. We performed user testing to discover the ease-of-use of our system as well as the subjective accuracy of its recommendations.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.