Please use this identifier to cite or link to this item:
Title: Enhangcing Collaborative Filtering Music recommendation by Balancing Exploration and Exploitation
Authors: XING ZHE
Keywords: music recommendation, collaborative filtering, reinforcement learning, Bayesian graphical model
Issue Date: 22-Aug-2014
Citation: XING ZHE (2014-08-22). Enhangcing Collaborative Filtering Music recommendation by Balancing Exploration and Exploitation. ScholarBank@NUS Repository.
Abstract: Collaborative filtering (CF) techniques have shown great success in music recommendation applications. However, traditional collaborative-filtering music recommendation algorithms work in a greedy way, invariably recommending songs with the highest predicted user ratings. Such a purely exploitative strategy may result in suboptimal performance over the long term. Using a reinforcement learning approach, we introduce exploration into CF and try to strike a balance between exploration and exploitation. In order to learn users' musical tastes, we use a Bayesian graphical model that takes account of both CF latent factors and recommendation novelty. Moreover, we designed a Bayesian inference algorithm to efficiently estimate the posterior rating distributions. To the best of our knowledge, this is the first attempt to remedy the greedy nature of CF approaches in music recommendation. Results from both simulation experiments and user study show that our proposed approach significantly improves music recommendation performance.
Appears in Collections:Master's Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
Thesis_XING ZHE_20141125.pdf1.19 MBAdobe PDF



Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.