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

The last decade has witnessed many successes of deep learning-based modelsfor industry-scale recommender systems. These models are typically trainedoffline in a batch manner. While being effective in capturing users' pastinteractions with recommendation platforms, batch learning suffers from longmodel-update latency and is vulnerable to system biases, making it hard toadapt to distribution shift and explore new items or user interests. Althoughonline learning-based approaches (e.g., multi-armed bandits) have demonstratedpromising theoretical results in tackling these challenges, their practicalreal-time implementation in large-scale recommender systems remains limited.First, the scalability of online approaches in servicing a massive onlinetraffic while ensuring timely updates of bandit parameters poses a significantchallenge. Additionally, exploring uncertainty in recommender systems caneasily result in unfavorable user experience, highlighting the need fordevising intricate strategies that effectively balance the trade-off betweenexploitation and exploration. In this paper, we introduce Online Matching: ascalable closed-loop bandit system learning from users' direct feedback onitems in real time. We present a hybrid "offline + online" approach forconstructing this system, accompanied by a comprehensive exposition of theend-to-end system architecture. We propose Diag-LinUCB -- a novel extension ofthe LinUCB algorithm -- to enable distributed updates of bandits parameter in ascalable and timely manner. We conduct live experiments in YouTube and showthat Online Matching is able to enhance the capabilities of fresh contentdiscovery and item exploration in the present platform.

Online Matching: A Real-time Bandit System for Large-scale Recommendations