Exposure Inequality in People Recommender Systems: The Long-Term Effects

Abstract

People recommender systems may affect the exposure that users receive in social networking platforms, influencing attention dynamics and potentially strengthening pre-existing inequalities that disproportionately affect certain groups. In this paper we introduce a model to simulate the feedback loop created by multiple rounds of interactions between users and a link recommender in a social network. This allows us to study the long-term consequences of those particular recommendation algorithms. Our model is equipped with several parameters to control: (i) the level of homophily in the network, (ii) the relative size of the groups, (iii) the choice among several state-of-the-art link recommenders, and (iv) the choice among three different stochastic user behavior models, that decide which recommendations are accepted or rejected. Our extensive experimentation with the proposed model shows that a minority group, if homophilic enough, can get a disproportionate advantage in exposure from all link recommenders. Instead, when it is heterophilic, it gets underexposed. Moreover, while the homophily level of the minority affects the speed of the growth of the disparate exposure, the relative size of the minority affects the magnitude of the effect. Finally, link recommenders strengthen exposure inequalities at the individual level, exacerbating the“rich-get-richer” effect: this happens for both the minority and the majority class and independently of their level of homophily.

Bias Assessment Approaches for Addressing User-Centered Fairness in GNN-Based Recommender Systems

“…The circle closes with the transfer of bias from the results back to the data. At each of these stages, new biases can be introduced [5,52]. This circular behavior can complicate the process of bias recognition and mitigation [5,53].…”

A Comparative Analysis of Bias Amplification in Graph Neural Network Approaches for Recommender Systems

“…This can lead to a neglection of the model toward unpopular items and give a higher score to the more popular ones [2]. Together, the previously mentioned biases can create a circle graph in which biased data move from one stage to the next, where additional and new biases are introduced [2,39]. This circular behavior of biases increases the complexity to recognize where actions are needed.…”

“…In the recommendation area, a model that simulates multiple rounds of a bias feedback loop in a social network was proposed in [39] in order to analyze the consequence of this feedback loop in the long run. This model uses different control parameters including the level of homophily in the network, the relative size of the groups, the choice among many new link recommenders, and the choice between three various stochastic use behavior models, which decide whether each recommendation would be accepted or not.…”

Gender Inversion Phenomenon in the Internet and Social Media

“…Previous research has focused chiefly on different genders' behavioral and treatment differences in social media. However, the phenomenon of gender inversion has received little attention [3,4]. The reason for focusing on this issue is that gender inversion is often overlooked but can reflect subconscious choices.…”