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

Recommender systems (RSs) employ user-item feedback, e.g., ratings, to matchcustomers to personalized lists of products. Approaches to top-k recommendationmainly rely on Learning-To-Rank algorithms and, among them, the most widelyadopted is Bayesian Personalized Ranking (BPR), which bases on a pair-wiseoptimization approach. Recently, BPR has been found vulnerable againstadversarial perturbations of its model parameters. Adversarial PersonalizedRanking (APR) mitigates this issue by robustifying BPR via an adversarialtraining procedure. The empirical improvements of APR's accuracy performance onBPR have led to its wide use in several recommender models. However, a keyoverlooked aspect has been the beyond-accuracy performance of APR, i.e.,novelty, coverage, and amplification of popularity bias, considering thatrecent results suggest that BPR, the building block of APR, is sensitive to theintensification of biases and reduction of recommendation novelty. In thiswork, we model the learning characteristics of the BPR and APR optimizationframeworks to give mathematical evidence that, when the feedback data have atailed distribution, APR amplifies the popularity bias more than BPR due to anunbalanced number of received positive updates from short-head items. Usingmatrix factorization (MF), we empirically validate the theoretical results byperforming preliminary experiments on two public datasets to compare BPR-MF andAPR-MF performance on accuracy and beyond-accuracy metrics. The experimentalresults consistently show the degradation of novelty and coverage measures anda worrying amplification of bias.

Understanding the Effects of Adversarial Personalized Ranking Optimization Method on Recommendation Quality