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

Ensuring the privacy of users whose data are used to train Natural LanguageProcessing (NLP) models is necessary to build and maintain customer trust.Differential Privacy (DP) has emerged as the most successful method to protectthe privacy of individuals. However, applying DP to the NLP domain comes withunique challenges. The most successful previous methods use a generalization ofDP for metric spaces, and apply the privatization by adding noise to inputs inthe metric space of word embeddings. However, these methods assume that onespecific distance measure is being used, ignore the density of the space aroundthe input, and assume the embeddings used have been trained on non-sensitivedata.  In this work we propose Truncated Exponential Mechanism (TEM), a generalmethod that allows the privatization of words using any distance metric, onembeddings that can be trained on sensitive data. Our method makes use of theexponential mechanism to turn the privatization step into a \emph{selectionproblem}. This allows the noise applied to be calibrated to the density of theembedding space around the input, and makes domain adaptation possible for theembeddings. In our experiments, we demonstrate that our method significantlyoutperforms the state-of-the-art in terms of utility for the same level ofprivacy, while providing more flexibility in the metric selection.

TEM: High Utility Metric Differential Privacy on Text