English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recommendation models trained on the user feedback collected from deployedrecommendation systems are commonly biased. User feedback is considerablyaffected by the exposure mechanism, as users only provide feedback on the itemsexposed to them and passively ignore the unexposed items, thus producingnumerous false negative samples. Inevitably, biases caused by such userfeedback are inherited by new models and amplified via feedback loops.Moreover, the presence of false negative samples makes negative samplingdifficult and introduces spurious information in the user preference modelingprocess of the model. Recent work has investigated the negative impact offeedback loops and unknown exposure mechanisms on recommendation quality anduser experience, essentially treating them as independent factors and ignoringtheir cross-effects. To address these issues, we deeply analyze the dataexposure mechanism from the perspective of data iteration and feedback loopswith the Missing Not At Random (\textbf{MNAR}) assumption, theoreticallydemonstrating the existence of an available stabilization factor in thetransformation of the exposure mechanism under the feedback loops. We furtherpropose Dynamic Personalized Ranking (\textbf{DPR}), an unbiased algorithm thatuses dynamic re-weighting to mitigate the cross-effects of exposure mechanismsand feedback loops without additional information. Furthermore, we design aplugin named Universal Anti-False Negative (\textbf{UFN}) to mitigate thenegative impact of the false negative problem. We demonstrate theoreticallythat our approach mitigates the negative effects of feedback loops and unknownexposure mechanisms. Experimental results on real-world datasets demonstratethat models using DPR can better handle bias accumulation and the universalityof UFN in mainstream loss methods.

DPR: An Algorithm Mitigate Bias Accumulation in Recommendation feedback loops