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

One major task of spoken language understanding (SLU) in modern personalassistants is to extract semantic concepts from an utterance, called slotfilling. Although existing slot filling models attempted to improve extractingnew concepts that are not seen in training data, the performance in practice isstill not satisfied. Recent research collected question and answer annotateddata to learn what is unknown and should be asked, yet not practically scalabledue to the heavy data collection effort. In this paper, we incorporatesoftmax-based slot filling neural architectures to model the sequenceuncertainty without question supervision. We design a Dirichlet Prior RNN tomodel high-order uncertainty by degenerating as softmax layer for RNN modeltraining. To further enhance the uncertainty modeling robustness, we propose anovel multi-task training to calibrate the Dirichlet concentration parameters.We collect unseen concepts to create two test datasets from SLU benchmarkdatasets Snips and ATIS. On these two and another existing Concept Learningbenchmark datasets, we show that our approach significantly outperformsstate-of-the-art approaches by up to 8.18%. Our method is generic and can beapplied to any RNN or Transformer based slot filling models with a softmaxlayer.

Modeling Token-level Uncertainty to Learn Unknown Concepts in SLU via Calibrated Dirichlet Prior RNN