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

Re-ranking draws increased attention on both academics and industries, whichrearranges the ranking list by modeling the mutual influence among items tobetter meet users' demands. Many existing re-ranking methods directly take theinitial ranking list as input, and generate the optimal permutation through awell-designed context-wise model, which brings the evaluation-before-rerankingproblem. Meanwhile, evaluating all candidate permutations brings unacceptablecomputational costs in practice. Thus, to better balance efficiency andeffectiveness, online systems usually use a two-stage architecture which usessome heuristic methods such as beam-search to generate a suitable amount ofcandidate permutations firstly, which are then fed into the evaluation model toget the optimal permutation. However, existing methods in both stages can beimproved through the following aspects. As for generation stage, heuristicmethods only use point-wise prediction scores and lack an effective judgment.As for evaluation stage, most existing context-wise evaluation models onlyconsider the item context and lack more fine-grained feature context modeling.This paper presents a novel end-to-end re-ranking framework named PIER totackle the above challenges which still follows the two-stage architecture andcontains two mainly modules named FPSM and OCPM. We apply SimHash in FPSM toselect top-K candidates from the full permutation based on user'spermutation-level interest in an efficient way. Then we design a novelomnidirectional attention mechanism in OCPM to capture the context informationin the permutation. Finally, we jointly train these two modules end-to-end byintroducing a comparative learning loss. Offline experiment results demonstratethat PIER outperforms baseline models on both public and industrial datasets,and we have successfully deployed PIER on Meituan food delivery platform.

PIER: Permutation-Level Interest-Based End-to-End Re-ranking Framework in E-commerce