Designing Recommender Systems to Depolarize

Stray, Jonathan

doi:10.48550/arxiv.2107.04953

Cited by 8 publications

(15 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Depending on this state, the model picks a particular video (left, center , right) as the action to take. 12 • The RL model goes back to the homepage and observes the recommendations to calculate the reward value using O.…”

Section: Training the Rl Modelmentioning

confidence: 99%

“…This may lead to a feedback loop-a potential vicious cycle of reinforcement. The worry is that this loop effect is prevalent on social media platforms and is a major contributing factor to polarization and radicalization 1 [11,12].…”

Section: Introductionmentioning

confidence: 99%

“…This allows the user to see personally interesting and relevant content. However, issues arise in the case of political content where these factors are feared to reinforce users' political biases, ultimately leading to radicalization, fomenting civil unrest, and endangering public health [11,12].…”

Section: Introductionmentioning

confidence: 99%

“…Note that the operator ∪ here denotes an ordered union between the set of videos, as the order in which videos are watched is important 12. The main motivation behind defining this action set is that each action should be able to send an explicit signal to the algorithm, to influence the recommendations.…”

mentioning

confidence: 99%

See 3 more Smart Citations

YouTube, The Great Radicalizer? Auditing and Mitigating Ideological Biases in YouTube Recommendations

Haroon¹,

Chhabra²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Recommendations algorithms of social media platforms are often criticized for placing users in "rabbit holes" of (increasingly) ideologically biased content. Despite these concerns, prior evidence on this algorithmic radicalization is inconsistent. Furthermore, prior work lacks systematic interventions that reduce the potential ideological bias in recommendation algorithms. We conduct a systematic audit of YouTube's recommendation system using a hundred thousand sock puppets to determine the presence of ideological bias (i.e., are recommendations aligned with users' ideology), its magnitude (i.e., are users recommended an increasing number of videos aligned with their ideology), and radicalization (i.e., are the recommendations progressively more extreme). Furthermore, we design and evaluate a bottomup intervention to minimize ideological bias in recommendations without relying on cooperation from YouTube. We find that YouTube's recommendations do direct users -especially right-leaning users -to ideologically biased and increasingly radical content on both homepages and in up-next recommendations. Our intervention effectively mitigates the observed bias, leading to more recommendations to ideologically neutral, diverse, and dissimilar content, yet debiasing is especially challenging for right-leaning users. Our systematic assessment shows that while YouTube recommendations lead to ideological bias, such bias can be mitigated through our intervention. * Additional information, including the accompanying source code and data, is available at: https://github.com/ucdavis-noyce/YouTube-Auditing-Mitigating-Bias 1 Naturally, most citizens do not inhabit insular communities online [7,8]. The small set that does, however, is socially consequential. Those users tend to be more partisan, more politically active, and have a disproportionate influence on the democratic process [9,10]

show abstract

Section: Training the Rl Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

YouTube, The Great Radicalizer? Auditing and Mitigating Ideological Biases in YouTube Recommendations

Haroon¹,

Chhabra²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…On the contrary, whatever mathematical precision is achieved, it will forever be off the mark as people intuitively anticipate how they are profiled, shift their expectations, and change their interactions. This also implies that “fixing” the Goodhart effect by way of better metrics, as some suggest (Stray, 2021 ), may instead exacerbate rather than solve the problem.…”

Section: Human-machine Feedback Loops and The Goodhart Effectmentioning

confidence: 99%

The Issue of Proxies and Choice Architectures. Why EU Law Matters for Recommender Systems

Hildebrandt

2022

Front. Artif. Intell.

View full text Add to dashboard Cite

Recommendations are meant to increase sales or ad revenue, as these are the first priority of those who pay for them. As recommender systems match their recommendations with inferred preferences, we should not be surprised if the algorithm optimizes for lucrative preferences and thus co-produces the preferences they mine. This relates to the well-known problems of feedback loops, filter bubbles, and echo chambers. In this article, I discuss the implications of the fact that computing systems necessarily work with proxies when inferring recommendations and raise a number of questions about whether recommender systems actually do what they are claimed to do, while also analysing the often-perverse economic incentive structures that have a major impact on relevant design decisions. Finally, I will explain how the choice architectures for data controllers and providers of AI systems as foreseen in the EU's General Data Protection Regulation (GDPR), the proposed EU Digital Services Act (DSA) and the proposed EU AI Act will help to break through various vicious circles, by constraining how people may be targeted (GDPR, DSA) and by requiring documented evidence of the robustness, resilience, reliability, and the responsible design and deployment of high-risk recommender systems (AI Act).

show abstract

A Comprehensive Survey of Detection and Prevention Approaches for Online Radicalization: Identifying Gaps and Future Directions

Berjawi,

Fenza,

Loia

2023

IEEE Access

View full text Add to dashboard Cite

In the digital era, online radicalization has emerged as a significant concern for governments, social media platforms, and researchers. Detecting and preventing online radicalization have become key priorities, leading to extensive research efforts. This study presents a comprehensive survey of existing works in this field, covering various techniques and methodologies. An extensive assessment of 68 publications from databases such as IEEE, SCOPUS, and Web of Science (WoS) was conducted to analyze recent literature on detecting and preventing online radicalization. This research provides an overview of the definition of online radicalization and its relationship with social media. It explores different types and sources of datasets used in studying online radicalization. Additionally, it categorizes approaches and techniques, including Machine Learning (ML), Deep Learning (DL), and Graph algorithms, for detecting and preventing online radicalization. The survey identifies limitations and challenges in the field, highlighting existing gaps and suggesting potential directions for further study. To the best of the authors' knowledge, this work is the first of its kind to undertake such a holistic investigation that consolidates these methodologies presenting them in an accessible manner. The findings contribute as a valuable resource for academics, decision-makers, and professionals working in the field of counter-radicalization and provide insights into existing countermeasures against this expanding threat.

show abstract

Designing Recommender Systems to Depolarize

Cited by 8 publications

References 0 publications

YouTube, The Great Radicalizer? Auditing and Mitigating Ideological Biases in YouTube Recommendations

YouTube, The Great Radicalizer? Auditing and Mitigating Ideological Biases in YouTube Recommendations

The Issue of Proxies and Choice Architectures. Why EU Law Matters for Recommender Systems

A Comprehensive Survey of Detection and Prevention Approaches for Online Radicalization: Identifying Gaps and Future Directions

Contact Info

Product

Resources

About