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

Life-long user behavior modeling, i.e., extracting a user's hidden interestsfrom rich historical behaviors in months or even years, plays a central role inmodern CTR prediction systems. Conventional algorithms mostly follow twocascading stages: a simple General Search Unit (GSU) for fast and coarse searchover tens of thousands of long-term behaviors and an Exact Search Unit (ESU)for effective Target Attention (TA) over the small number of finalists fromGSU. Although efficient, existing algorithms mostly suffer from a cruciallimitation: the \textit{inconsistent} target-behavior relevance metrics betweenGSU and ESU. As a result, their GSU usually misses highly relevant behaviorsbut retrieves ones considered irrelevant by ESU. In such case, the TA in ESU,no matter how attention is allocated, mostly deviates from the real userinterests and thus degrades the overall CTR prediction accuracy. To addresssuch inconsistency, we propose \textbf{TWo-stage Interest Network (TWIN)},where our Consistency-Preserved GSU (CP-GSU) adopts the identicaltarget-behavior relevance metric as the TA in ESU, making the two stages twins.Specifically, to break TA's computational bottleneck and extend it from ESU toGSU, or namely from behavior length $10^2$ to length $10^4-10^5$, we build anovel attention mechanism by behavior feature splitting. For the video inherentfeatures of a behavior, we calculate their linear projection by efficientpre-computing \& caching strategies. And for the user-item cross features, wecompress each into a one-dimentional bias term in the attention scorecalculation to save the computational cost. The consistency between two stages,together with the effective TA-based relevance metric in CP-GSU, contributes tosignificant performance gain in CTR prediction.

TWIN: TWo-stage Interest Network for Lifelong User Behavior Modeling in CTR Prediction at Kuaishou