Music recommendation model based on user listening behavior and utility based preference scoring
Date of Publication
Master of Science in Computer Science
College of Computer Studies
Defense Panel Chair
Merlin Teodosia Suarez
Defense Panel Member
Recommending the most appropriate music is one of the most studied fields in the contest of Music Information Retrieval. Music Recommendation modules often take note of the users music preferences when it comes to recommending music. In this study, approaches such as Music Similarity, have also been applied during the recommendation phase.
The study made use of normalized acoustic features extracted using MIR tools MARSYAS 0.5.0 alpha 1 and Audio 1.0.4 and utility based preference scoring to find relevant music to be used as recommendations. Using this approach, the study was able to come up with an average True-Positive rating of 54.43% in determining the songs the user will select for the month given previous months data.
This study made use of a recommendation formula that can be used for future studies. Some examples could be a different set of similarity measures used, more computational functions to use as a basis for recommendation, as well as changing constant values used throughout the computational functions used during the research. Applying suggestions for measuring utility can also be used for further studies who wish to go into dynamic and more active recommendation models.
Archives, The Learning Commons, 12F Henry Sy Sr. Hall
leaves ; 4 3/4 in.
Caronongan, A. (2014). Music recommendation model based on user listening behavior and utility based preference scoring. Retrieved from https://animorepository.dlsu.edu.ph/etd_masteral/4627