It Is Different When Items Are Older

Huang, Jin; Oosterhuis, Harrie; Rijke, Maarten de

doi:10.1145/3488560.3498375

Cited by 28 publications

(5 citation statements)

References 44 publications

(41 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One paper in the area of proving bias changes over time is 'It Is Different When Items Are Older: Debiasing Recommendations When Selection Bias and User Preferences Are Dynamic'. This paper demonstrates that in the real world, time-aware methods can predict selection bias better than static methods [1]. However, it does not consider biases that depend on the previous bias.…”

Section: Introductionmentioning

confidence: 95%

See 1 more Smart Citation

Improved time-aware debiasing models for recommender systems

Wang,

Wu,

Zhao

et al. 2023

ACE

View full text Add to dashboard Cite

Recommender systems are affected by user selection bias during user interactions. Previously, Huang, Jin et al. proposed the TMF-DANCER model to deal with the fact that selection bias is dynamic, and the popularity of an item and user preferences may change drastically over time. However, the previous time-aware methods did not consider the continuity of time, and performing collaborative filtering with matrix factorization may not be able to capture the complex structure of user interaction data. Our methods enhance the effectiveness of time-aware debiasing by utilizing the time-varying sequential method to alleviate the discontinuity of time-based on the frequency-based method, Replacing the matrix factorization with neural collaborative filtering (NCF). We used movie lens 100k as the data for training. The experimental results indicate that methods incorporating the time-varying sequential method and NCF have better performance in rating prediction than debiasing methods without.

show abstract

Section: Introductionmentioning

confidence: 95%

“…Huang et al proposed eight different methods to estimate propensity in a time-varying scenario [1]. They use 𝜎 to denotate the sigmoid function.…”

Section: Propensity Estimation With the Time Effectmentioning

confidence: 99%

Improved time-aware debiasing models for recommender systems

Wang,

Wu,

Zhao

et al. 2023

ACE

View full text Add to dashboard Cite

show abstract

“…[40] models the dependencies between items inside a session and re-ranking works also focus on the pair-wise or list-wise item-item relationship constructions [1,26,27]. Work [14] focuses on the age of an item and its effect on selection bias and user preferences for enhancements in rating predictions and work [38] uses survival analysis in recommended opportunities modelings for accuracy enhancements in e-commerce scenarios. Survival analysis, also called time-to-event analysis [15,16,39], is a branch of techniques that focused on lifetime modelings, such as the death in biological organisms, failure in mechanical systems, and user churns in games [18].…”

Section: Item Timeliness Modelingmentioning

confidence: 99%

Measuring Item Global Residual Value for Fair Recommendation

Wang

Jiang

et al. 2023

Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

In the era of information explosion, numerous items emerge every day, especially in feed scenarios. Due to the limited system display slots and user browsing attention, various recommendation systems are designed not only to satisfy users' personalized information needs but also to allocate items' exposure. However, recent recommendation studies mainly focus on modeling user preferences to present satisfying results and maximize user interactions, while paying little attention to developing item-side fair exposure mechanisms for rational information delivery. This may lead to serious resource allocation problems on the item side, such as the Snowball Effect. Furthermore, unfair exposure mechanisms may hurt recommendation performance. In this paper, we call for a shift of attention from modeling user preferences to developing fair exposure mechanisms for items. We first conduct empirical analyses of feed scenarios to explore exposure problems between items with distinct uploaded times. This points out that unfair exposure caused by the time factor may be the major cause of the Snowball Effect. Then, we propose to explicitly model item-level customized timeliness distribution, Global Residual Value (GRV), for fair resource allocation. This GRV module is introduced into recommendations with the designed Timeliness-aware Fair Recommendation Framework (TaFR). Extensive experiments on two datasets demonstrate that TaFR achieves consistent improvements with various backbone recommendation models. By modeling item-side customized Global Residual Value, we achieve a fairer distribution of resources and, at the same time, improve recommendation performance. CCS CONCEPTS• Information systems → Recommender systems.

show abstract

“…What can be considered a special case of re-weighting are methods based on Inverse Propensity Scoring (IPS) [69,92,118]. The concept of inverse propensity has been adopted from statistics and utilized in several prior works to reduce the influence of popularity.…”

Section: Bias Mitigation Approachesmentioning

confidence: 99%

“…Additionally, this work utilizes causal inference and counterfactual reasoning for unbiased recommendation quality estimation. Yang et al [69] later described a related approach which additionally considers the dynamic aspect of propensity scoring. The authors argue that recommendation algorithms should account for user preference changes over time.…”

Section: Bias Mitigation Approachesmentioning

confidence: 99%

Mitigating Popularity Bias in Recommendation: Potential and Limits of Calibration Approaches

Klimashevskaia

Elahi

Jannach

et al. 2022

Communications in Computer and Information Science

View full text Add to dashboard Cite

Recommender systems help people find relevant content in a personalized way. One main promise of such systems is that they are able to increase the visibility of items in the long tail, i.e., the lesserknown items in a catalogue. Existing research, however, suggests that in many situations today's recommendation algorithms instead exhibit a popularity bias, meaning that they often focus on rather popular items in their recommendations. Such a bias may not only lead to limited value of the recommendations for consumers and providers in the short run, but it may also cause undesired reinforcement effects over time. In this paper, we discuss the potential reasons for popularity bias and we review existing approaches to detect, quantify and mitigate popularity bias in recommender systems. Our survey therefore includes both an overview of the computational metrics used in the literature as well as a review of the main technical approaches to reduce the bias. We furthermore critically discuss today's literature, where we observe that the research is almost entirely based on computational experiments and on certain assumptions regarding the practical effects of including long-tail items in the recommendations.

show abstract

It Is Different When Items Are Older

Cited by 28 publications

References 44 publications

Improved time-aware debiasing models for recommender systems

Improved time-aware debiasing models for recommender systems

Measuring Item Global Residual Value for Fair Recommendation

Mitigating Popularity Bias in Recommendation: Potential and Limits of Calibration Approaches

Contact Info

Product

Resources

About