Recent task-oriented dialog systems have had great success in building English-based personal assistants, but extending these systems to a global audience is challenging due to the need for annotated data in the target language. An alternative approach is to leverage existing data in a high-resource language to enable cross-lingual transfer in low-resource language models. However, this type of transfer has not been widely explored in natural language response generation. In this research, we investigate the use of state-of-the-art multilingual models such as mBART and T5 to facilitate zero-shot and few-shot transfer of code-switched responses. We propose a new adapter-based framework that allows for efficient transfer by learning task-specific representations and encapsulating source and target language representations. Our framework is able to successfully transfer language knowledge even when the target language corpus is limited. We present both quantitative and qualitative analyses to evaluate the effectiveness of our approach.

In multi-turn dialog understanding, semantic frames are constructed by detecting intents and slots within each user utterance. However, recent works lack the capability of modeling multi-turn dynamics within a dialog in natural language understanding (NLU), instead leaving them for updating dialog states only. Moreover, humans usually associate relevant background knowledge with the current dialog contexts to better illustrate slot semantics revealed from word connotations, where previous works have explored such possibility mostly in knowledge-grounded response generation. In this paper, we propose to amend the research gap by equipping a BERT-based NLU framework with knowledge and context awareness. We first encode dialog contexts with a unidirectional context-aware transformer encoder and select relevant inter-word knowledge with the current word and previous history based on a knowledge attention mechanism. Experimental results in two complicated multi-turn dialog datasets have demonstrated significant improvements of our proposed framework. Attention visualization also demonstrates how our modules leverage knowledge across the utterance.

With the early success of query-answer assistants such as Alexa and Siri, research attempts to expand system capabilities of handling service automation are now abundant. However, preliminary systems have quickly found the inadequacy in relying on simple classification techniques to effectively accomplish the automation task. The main challenge is that the dialogue often involves complexity in user’s intents (or purposes) which are multiproned, subject to spontaneous change, and difficult to track. Furthermore, public datasets have not considered these complications and the general semantic annotations are lacking which may result in zero-shot problem. Motivated by the above, we propose a Label-Aware BERT Attention Network (LABAN) for zero-shot multi-intent detection. We first encode input utterances with BERT and construct a label embedded space by considering embedded semantics in intent labels. An input utterance is then classified based on its projection weights on each intent embedding in this embedded space. We show that it successfully extends to few/zero-shot setting where part of intent labels are unseen in training data, by also taking account of semantics in these unseen intent labels. Experimental results show that our approach is capable of detecting many unseen intent labels correctly. It also achieves the state-of-the-art performance on five multi-intent datasets in normal cases.