Leveraging Position Bias to Improve Peer Recommendation

By Kristina Lerman, Tad Hogg. 2014

110 citations recorded by the source corpus.

With the advent of social media and peer production, the amount of new online content has grown dramatically. To identify interesting items in the vast stream of new content, providers must rely on peer recommendation to aggregate opinions of their many users. Due to human cognitive biases, the presentation order strongly affects how people allocate attention to the available content. Moreover, we can manipulate attention through the presentation order of items to change the way peer recommendation works. We experimentally evaluate this effect using Amazon Mechanical Turk. We find that different policies for ordering content can steer user attention so as to improve the outcomes of peer recommendation.

Subfields: Computer Science Applications, Information Systems, Sociology and Political Science, Recommender Systems, AI Ethics, Fairness & Society

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Science Applications

In Information Systems

In Sociology and Political Science

In Recommender Systems

In AI Ethics, Fairness & Society

Citations in the corpus

Social Data: Biases, Methodological Pitfalls, and Ethical Boundaries (2019, background): “Biases that bound the applicability of general solutions may help boost the performance of dedicated solutions (Yan et al., 2011; Olteanu et al., 2014a) or may inform their design ( Olteanu and Pierre, 2012; Lerman and Hogg, 2014).”
Information is not a Virus, and Other Consequences of Human Cognitive Limits (2016, methodology): “Researchers conducted controlled experiments on Amazon Mechanical Turk to quantify position bias [36].”
The Benefits of Social Influence in Optimized Cultural Markets (2015, result): “The performance results also shed light on recent results that demonstrated the benefits of position bias in the absence of social influence [18].”
Towards a Standard for Identifying and Managing Bias in Artificial Intelligence (2022, unknown): “The idea that top-ranked results are the most relevant and important and will result in more clicks than other results [133, 283].”
The myopia of crowds: Cognitive load and collective evaluation of answers on Stack Exchange (2016, unknown): “We know from previous research that people’s choices are biased by the rank order of items [8, 21, 35].”
Aligning Popularity and Quality in Online Cultural Markets (2021, background): “Social influence is also reinforced by position bias (e.g., (Lerman and Hogg 2014)), as consumer preferences are also affected considerably by the visibility of the choices.”
Taming the Unpredictability of Cultural Markets with Social Influence (2017, background): “Specifically, ordering items by recency of recommendation decreased the inequality and increased predictability of collective outcomes of peer recommendation [16].”
Disentangling the Effects of Social Signals (2014, methodology): “This quantity decreases rapidly with position: a story at the top of a list gets about five times as much attention as a story in the middle of the list, consistent with direct measurements reported in (Lerman and Hogg, 2014).”