Listening factors: a large-scale principal components analysis of long-term music listening histories
Listening factors: a large-scale principal components analysis of long-term music listening histories
复制标题
聆听因素：长期音乐聆听历史的大规模主成分分析
DOI：
10.1145/2207676.2208581
复制
发表时间：
2012
期刊：
影响因子：
--
通讯作者：
A. Butz
中科院分区：
文献类型：
--
作者：
D. Baur;Jennifer Büttgen;A. Butz
有多少音乐爱好者，就有多少听音乐的策略。这使得学习总体模式和相似性变得困难。在本文中，我们提出了一个长期的音乐听历史的实证分析，从last.fm网络服务。它让我们深入了解音乐聆听行为中最显著的因素。我们的样本包含310个历史记录，持续时间长达6年，48个相关变量描述了各种用户和音乐特征。使用主成分分析，我们将这些变量汇总为13个成分，并发现它们之间的几个相关性。分析特别显示了季节和听众对新奇音乐的兴趣对音乐选择的影响。使用这些信息，用户收听历史的样本甚至只是人口统计数据都可以用来创建个性化界面和新颖的推荐策略。最后，我们得出未来音乐界面的设计考虑。
There are about as many strategies for listening to music as there are music enthusiasts. This makes learning about overarching patterns and similarities difficult. In this paper, we present an empirical analysis of long-term music listening histories from the last.fm web service. It gives insight into the most distinguishing factors in music listening behavior. Our sample contains 310 histories with up to six years duration and 48 associated variables describing various user and music characteristics. Using a principal components analysis, we aggregated these variables into 13 components and found several correlations between them. The analysis especially showed the impact of seasons and a listener's interest in novelty on music choice. Using this information, a sample of a user's listening history or even just demographical data could be used to create personalized interfaces and novel recommendation strategies. We close with derived design considerations for future music interfaces.