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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Simultaneous machine translation (SiMT) starts translating while receivingthe streaming source inputs, and hence the source sentence is always incompleteduring translating. Different from the full-sentence MT using the conventionalseq-to-seq architecture, SiMT often applies prefix-to-prefix architecture,which forces each target word to only align with a partial source prefix toadapt to the incomplete source in streaming inputs. However, the source wordsin the front positions are always illusoryly considered more important sincethey appear in more prefixes, resulting in position bias, which makes the modelpay more attention on the front source positions in testing. In this paper, wefirst analyze the phenomenon of position bias in SiMT, and develop aLength-Aware Framework to reduce the position bias by bridging the structuralgap between SiMT and full-sentence MT. Specifically, given the streaminginputs, we first predict the full-sentence length and then fill the futuresource position with positional encoding, thereby turning the streaming inputsinto a pseudo full-sentence. The proposed framework can be integrated into mostexisting SiMT methods to further improve performance. Experiments on tworepresentative SiMT methods, including the state-of-the-art adaptive policy,show that our method successfully reduces the position bias and therebyachieves better SiMT performance.

Reducing Position Bias in Simultaneous Machine Translation with Length-Aware Framework