Modeling Popularity and Temporal Drift of Music Genre Preferences

Abstract

In this paper, we address the problem of modeling and predicting the music genre preferences of users. We introduce a novel user modeling approach, BLL u , which takes into account the popularity of music genres as well as temporal drifts of user listening behavior. To model these two factors, BLL u adopts a psychological model that describes how humans access information in their memory. We evaluate our approach on a standard dataset of Last.fm listening histories, which contains fine-grained music genre information. To investigate performance for different types of users, we assign each user a mainstreaminess value that corresponds to the distance between the user's music genre preferences and the music genre preferences of the (Last.fm) mainstream. We adopt BLL u to model the listening habits and to predict the music genre preferences of three user groups: listeners of (i) niche, low-mainstream music, (ii) mainstream music, and (iii) medium-mainstream music that lies in-between. Our results show that BLL u provides the highest accuracy for predicting music genre preferences, compared to five baselines: (i) group-based modeling, (ii) user-based collaborative filtering, (iii) item-based collaborative filtering, (iv) frequency-based modeling, and (v) recencybased modeling. Besides, we achieve the most substantial accuracy improvements for the low-mainstream group. We believe that our findings provide valuable insights into the design of music recommender systems.

Support the underground: characteristics of beyond-mainstream music listeners

“…The presented approach helps add music from the long-tail into the recommendation list. In our previous research [41,42], we have used a framework [43] that employs insights from human memory theory to design a music recommendation algorithm that provides more accurate recommendations than collaborative filtering-based approaches for three groups of users, i.e., low-mainstream, mediummainstream and high-mainstream users. While the awareness of popularity bias in music recommender systems increases (e.g., [44]), the characteristics of music consumers whose preferences lie beyond popular, mainstream music are still not well understood.…”

Analyzing Item Popularity Bias of Music Recommender Systems: Are Different Genders Equally Affected?

“…We focuse on popularity bias, a well-studied form of bias in recommender systems research. This form of bias refers to the underrepresentation of less popular items in the produced recommendations and can lead to a significantly worse recommendation quality for consumers of long tail or niche items [3,10,12,13]. Abdollahpouri et al [3] show that state-of-the-art movie recommendation algorithms suffer from popularity bias, and introduce the delta-GAP metric to quantify the level of underrepresentation.…”

Predicting Music Relistening Behavior Using the ACT-R Framework

“…Algorithm Configurations. To configure the decay parameter of the Base-level component and in line with previous studies [12], we analyze the relistening behavior using a log-log plot as shown in Figure 1. We observe a good fit for a power law distribution of relistening behavior, i.e., linear fit in the log-log space.…”

“…The declarative memory module of ACT-R stores and retrieves information and consists of separate components. In contrast to prior work, which uses the base-level component to model predict user preferences in various domains (e.g., [9][10][11][12]14]), we investigate the utility of five components, i.e., (i) base-level, (ii) spreading, (iii) partial matching, (iv) valuation, and (v) noise component, in a next track prediction scenario. These components enable us to investigate the effect of various factors on track relistening behavior, i.e., (i) recency and frequency of prior exposure to tracks, (ii) co-occurrence of tracks, (iii) the similarity between tracks, (iv) familiarity with tracks, and (v) randomness in behavior.…”