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

While NMT has achieved remarkable results in the last 5 years, productionsystems come with strict quality requirements in arbitrarily niche domains thatare not always adequately covered by readily available parallel corpora. Thisis typically addressed by training domain specific models, using fine-tuningmethods and some variation of back-translation on top of in-domain monolingualcorpora. However, industrial practitioners can rarely afford to focus on asingle domain. A far more typical scenario includes a set of closely related,yet succinctly different sub-domains. At Booking.com, we need to translateproperty descriptions, user reviews, as well as messages, (for example thosesent between a customer and an agent or property manager). An editor might needto translate articles across a set of different topics. An e-commerce platformwould typically need to translate both the description of each item and theuser generated content related to them. To this end, we propose MDT: a novelmethod to simultaneously fine-tune on several sub-domains by passingmultidimensional sentence-level information to the model during training andinference. We show that MDT achieves results competitive to N specialist modelseach fine-tuned on a single constituent domain, while effectively serving all Nsub-domains, therefore cutting development and maintenance costs by the samefactor. Besides BLEU (industry standard automatic evaluation metric known toonly weakly correlate with human judgement) we also report rigorous humanevaluation results for all models and sub-domains as well as specific examplesthat better contextualise the performance of each model in terms of adequacyand fluency. To facilitate further research, we plan to make the code availableupon acceptance.

Multi-Domain Adaptation in Neural Machine Translation Through Multidimensional Tagging