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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

A primary challenge in ITE estimation is sample selection bias. Traditionalapproaches utilize treatment regularization techniques such as the IntegralProbability Metrics (IPM), re-weighting, and propensity score modeling tomitigate this bias. However, these regularizations may introduce undesirableinformation loss and limit the performance of the model. Furthermore, treatmenteffects vary across different external contexts, and the existing methods areinsufficient in fully interacting with and utilizing these contextual features.To address these issues, we propose a Context-Aware uplift model based on theTwo-Stage training approach (TSCAN), comprising CAN-U and CAN-D sub-models. Inthe first stage, we train an uplift model, called CAN-U, which includes thetreatment regularizations of IPM and propensity score prediction, to generate acomplete dataset with counterfactual uplift labels. In the second stage, wetrain a model named CAN-D, which utilizes an isotonic output layer to directlymodel uplift effects, thereby eliminating the reliance on the regularizationcomponents. CAN-D adaptively corrects the errors estimated by CAN-U throughreinforcing the factual samples, while avoiding the negative impacts associatedwith the aforementioned regularizations. Additionally, we introduce aContext-Aware Attention Layer throughout the two-stage process to manage theinteractions between treatment, merchant, and contextual features, therebymodeling the varying treatment effect in different contexts. We conductextensive experiments on two real-world datasets to validate the effectivenessof TSCAN. Ultimately, the deployment of our model for real-world merchantdiagnosis on one of China's largest online food ordering platforms validatesits practical utility and impact.

TSCAN: Context-Aware Uplift Modeling via Two-Stage Training for Online Merchant Business Diagnosis