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

Collaborative Filtering (CF) recommender models highly depend on user-iteminteractions to learn CF representations, thus falling short of recommendingcold-start items. To address this issue, prior studies mainly introduce itemfeatures (e.g., thumbnails) for cold-start item recommendation. They learn afeature extractor on warm-start items to align feature representations withinteractions, and then leverage the feature extractor to extract the featurerepresentations of cold-start items for interaction prediction. Unfortunately,the features of cold-start items, especially the popular ones, tend to divergefrom those of warm-start ones due to temporal feature shifts, preventing thefeature extractor from accurately learning feature representations ofcold-start items.  To alleviate the impact of temporal feature shifts, we consider usingDistributionally Robust Optimization (DRO) to enhance the generation ability ofthe feature extractor. Nonetheless, existing DRO methods face an inconsistencyissue: the worse-case warm-start items emphasized during DRO training might notalign well with the cold-start item distribution. To capture the temporalfeature shifts and combat this inconsistency issue, we propose a novel temporalDRO with new optimization objectives, namely, 1) to integrate a worst-casefactor to improve the worst-case performance, and 2) to devise a shiftingfactor to capture the shifting trend of item features and enhance theoptimization of the potentially popular groups in cold-start items. Substantialexperiments on three real-world datasets validate the superiority of ourtemporal DRO in enhancing the generalization ability of cold-start recommendermodels. The code is available at https://github.com/Linxyhaha/TDRO/.

Temporally and Distributionally Robust Optimization for Cold-start Recommendation