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/

Zendy Plus

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

Zendy Tools

Zendy Open

Existing advertisements click-through rate (CTR) prediction models are mainlydependent on behavior ID features, which are learned based on the historicaluser-ad interactions. Nevertheless, behavior ID features relying on historicaluser behaviors are not feasible to describe new ads without previousinteractions with users. To overcome the limitations of behavior ID features inmodeling new ads, we exploit the visual content in ads to boost the performanceof CTR prediction models. Specifically, we map each ad into a set of visual IDsbased on its visual content. These visual IDs are further used for generatingthe visual embedding for enhancing CTR prediction models. We formulate thelearning of visual IDs into a supervised quantization problem. Due to a lack ofclass labels for commercial images in advertisements, we exploit image textualdescriptions as the supervision to optimize the image extractor for generatingeffective visual IDs. Meanwhile, since the hard quantization isnon-differentiable, we soften the quantization operation to make it support theend-to-end network training. After mapping each image into visual IDs, we learnthe embedding for each visual ID based on the historical user-ad interactionsaccumulated in the past. Since the visual ID embedding depends only on thevisual content, it generalizes well to new ads. Meanwhile, the visual IDembedding complements the ad behavior ID embedding. Thus, it can considerablyboost the performance of the CTR prediction models previously relying onbehavior ID features for both new ads and ads that have accumulated rich userbehaviors. After incorporating the visual ID embedding in the CTR predictionmodel of Baidu online advertising, the average CTR of ads improves by 1.46%,and the total charge increases by 1.10%.

Boost CTR Prediction for New Advertisements via Modeling Visual Content