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

In e-commerce search, relevance between query and documents is an essentialrequirement for satisfying user experience. Different from traditionale-commerce platforms that offer products, users search on life serviceplatforms such as Meituan mainly for product providers, which usually haveabundant structured information, e.g. name, address, category, thousands ofproducts. Modeling search relevance with these rich structured contents ischallenging due to the following issues: (1) there is language distributiondiscrepancy among different fields of structured document, making it difficultto directly adopt off-the-shelf pretrained language model based methods likeBERT. (2) different fields usually have different importance and their lengthvary greatly, making it difficult to extract document information helpful forrelevance matching.  To tackle these issues, in this paper we propose a novel two-stagepretraining and matching architecture for relevance matching with richstructured documents. At pretraining stage, we propose an effective pretrainingmethod that employs both query and multiple fields of document as inputs,including an effective information compression method for lengthy fields. Atrelevance matching stage, a novel matching method is proposed by leveragingdomain knowledge in search query to generate more effective documentrepresentations for relevance scoring. Extensive offline experiments and onlineA/B tests on millions of users verify that the proposed architectureseffectively improve the performance of relevance modeling. The model hasalready been deployed online, serving the search traffic of Meituan for over ayear.

SPM: Structured Pretraining and Matching Architectures for Relevance Modeling in Meituan Search