Cross-Lingual Approaches to Reference Resolution in Dialogue Systems

  • 2018-11-27 18:52:58
  • Amr Sharaf, Arpit Gupta, Hancheng Ge, Chetan Naik, Lambert Mathias
  • 4


In the slot-filling paradigm, where a user can refer back to slots in thecontext during the conversation, the goal of the contextual understandingsystem is to resolve the referring expressions to the appropriate slots in thecontext. In this paper, we build on the context carryoversystem~\citep{Naik2018ContextualSC}, which provides a scalable multi-domainframework for resolving references. However, scaling this approach acrosslanguages is not a trivial task, due to the large demand on acquisition ofannotated data in the target language. Our main focus is on cross-lingualmethods for reference resolution as a way to alleviate the need for annotateddata in the target language. In the cross-lingual setup, we assume there isaccess to annotated resources as well as a well trained model in the sourcelanguage and little to no annotated data in the target language. In this paper,we explore three different approaches for cross-lingual transfer \textemdash~\delexicalization as data augmentation, multilingual embeddings and machinetranslation. We compare these approaches both on a low resource setting as wellas a large resource setting. Our experiments show that multilingual embeddingsand delexicalization via data augmentation have a significant impact in the lowresource setting, but the gains diminish as the amount of available data in thetarget language increases. Furthermore, when combined with machine translationwe can get performance very close to actual live data in the target language,with only 25\% of the data projected into the target language.


Introduction (beta)



Conclusion (beta)