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

We consider the problem of retrieving and ranking items in an eCommercecatalog, often called SKUs, in order of relevance to a user-issued query. Theinput data for the ranking are the texts of the queries and textual fields ofthe SKUs indexed in the catalog. We review the ways in which this problem bothresembles and differs from the problems of IR in the context of web search. Thedifferences between the product-search problem and the IR problem of web searchnecessitate a different approach in terms of both models and datasets. We firstreview the recent state-of-the-art models for web search IR, distinguishingbetween two distinct types of model which we call the distributed type and thelocal-interaction type. The different types of relevance models developed forIR have complementary advantages and disadvantages when applied to eCommerceproduct search. Further, we explain why the conventional methods for datasetconstruction employed in the IR literature fail to produce data which sufficesfor training or evaluation of models for eCommerce product search. We explainhow our own approach, applying task modeling techniques to the click-throughlogs of an eCommerce site, enables the construction of a large-scale datasetfor training and robust benchmarking of relevance models. Our experimentsconsist of applying several of the models from the IR literature to our owndataset. Empirically, we have established that, when applied to our dataset,certain models of local-interaction type reduce ranking errors by one-thirdcompared to the baseline tf-idf. Applied to our dataset, the distributed modelsfail to outperform the baseline. As a basis for a deployed system, thedistributed models have several advantages, computationally, over thelocal-interaction models. This motivates an ongoing program of work, which weoutline at the conclusion of the paper.

End-to-End Neural Ranking for eCommerce Product Search: an application of task models and textual embeddings