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 present a novel online algorithm that learns the essence of each dimensionin word embeddings by minimizing the within-group distance of contextualizedembedding groups. Three state-of-the-art neural-based language models are used,Flair, ELMo, and BERT, to generate contextualized word embeddings such thatdifferent embeddings are generated for the same word type, which are grouped bytheir senses manually annotated in the SemCor dataset. We hypothesize that notall dimensions are equally important for downstream tasks so that our algorithmcan detect unessential dimensions and discard them without hurting theperformance. To verify this hypothesis, we first mask dimensions determinedunessential by our algorithm, apply the masked word embeddings to a word sensedisambiguation task (WSD), and compare its performance against the one achievedby the original embeddings. Several KNN approaches are experimented toestablish strong baselines for WSD. Our results show that the masked wordembeddings do not hurt the performance and can improve it by 3%. Our work canbe used to conduct future research on the interpretability of contextualizedembeddings.

Incremental Sense Weight Training for the Interpretation of Contextualized Word Embeddings