Quantitative Study of Music Listening Behavior in a Social and Affective Context

LJ2M dataset: Toward better understanding of music listening behavior and user mood

“…Yang and Liu have used a smaller LiveJournal dataset LJ40k for a quantitative evaluation, comparing content-and context-based approaches to predict user mood [12]. In contrast, the current work presents a qualitative study on LJ2M.…”

“…Specifically, we use the mood entries to represent the affective context of users and employ music emotion recognition (MER) techniques [12] to predict the music emotion of the selected music. We note that, while a great deal of research has been made for MER in the MIR community [20,21], much less efforts have been devoted to studying the influence of user mood on music listening behavior, mostly due to the absence of a dataset that contains user mood information.…”

“…• These posts contain 88,164 unique song titles of 12,201 artists. For most songs, short audio previews (mostly 30-seconds) can be retrieved from 7digital.…”

“…From a scientific point of view, LJ2M enables deeper understanding of the interrelationship among the personal, situational, and musical factors of music listening [11]. For applications, LJ2M can be employed for studying contextaware music recommendation [3,4], emotion-based playlist generation [12], affective human-computer interface [13], and music therapy [14], to name a few.…”

“…As an example application, we present research that investigates the interaction between the affective context of the listener and the affective content of music, by using a stateof-the-art music emotion recognition system [12] for analyzing affective content of the audio previews, an affect lexicon called ANEW [15] for emotion visualization, and the psycholinguistic features computed by LIWC for text analysis of the blog articles [10]. The study offers insights into the role of music in mood regulation [16] and demonstrates how LJ2M can be used for studying real-world music listening behaviors.…”