End-to-End Slot Alignment and Recognition for Cross-Lingual NLU

• DOI: 10.18653/v1/2020.emnlp-main.410
• Publication Date: 2020-11-29T14:51:46Z
• Authors: Weijia Xu; Batool Haider; Saab Mansour
• Keywords: Natural language understanding

Natural language understanding (NLU) in the context of goal-oriented dialog systems typically includes intent classification and slot labeling tasks. Existing methods to expand an NLU system new languages use machine translation with label projection from source translated utterances, thus are sensitive errors. In this work, we propose a novel end-to-end model that learns align predict target labels jointly for cross-lingual transfer. We introduce MultiATIS++, multilingual corpus extends Multilingual ATIS nine across four families, evaluate our method using corpus. Results show outperforms simple fast-align on most languages, achieves competitive performance more complex, state-of-the-art only half training time. release MultiATIS++ community continue future research NLU.