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 system based on historical user-item interactions is of vitalimportance for web-based services. However, the observed data used to train therecommender model suffers from severe bias issues. Practically, the itemfrequency distribution of the dataset is a highly skewed power-lawdistribution. Interactions of a small fraction of head items account for almostthe whole training data. The normal training paradigm from such biased datatends to repetitively generate recommendations from the head items, whichfurther exacerbates the biases and affects the exploration of potentiallyinteresting items from the niche set. In this work, we innovatively explore thecentral theme of recommendation debiasing from an item cluster-wisemulti-objective optimization perspective. Aiming to balance the learning onvarious item clusters that differ in popularity during the training process, wepropose a model-agnostic framework namely Item Cluster-Wise Pareto-EfficientRecommendation (ICPE). In detail, we define our item cluster-wise optimizationtarget as the recommender model should balance all item clusters that differ inpopularity, thus we set the model learning on each item cluster as a uniqueoptimization objective. To achieve this goal, we first explore items'popularity levels from a novel causal reasoning perspective. Then, we devisepopularity discrepancy-based bisecting clustering to separate the itemclusters. Next, we adaptively find the overall harmonious gradient directionfor cluster-wise optimization objectives from a Pareto-efficient solver.Finally, in the prediction stage, we perform counterfactual inference tofurther eliminate the impact of global propensity. Extensive experimentalresults verify the superiorities of ICPE on overall recommendation performanceand biases elimination.

ICPE: An Item Cluster-Wise Pareto-Efficient Framework for Recommendation Debiasing