Data Augmentation with Atomic Templates for Spoken Language Understanding

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 3637–3643, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 3637 Data Augmentation with Atomic Templates for Spoken Language Understanding Zijian Zhao∗, Su Zhu∗and Kai Yu† MoE Key Lab of Artificial Intelligence SpeechLab, Department of Computer Science and Engineering Shanghai Jiao Tong University, Shanghai, China {1248uu,paul2204,kai.yu}@sjtu.edu.cn Abstract Spoken Language Understanding (SLU) con- verts user utterances into structured seman- tic representations. Data sparsity is one of the main obstacles of SLU due to the high cost of human annotation, especially when do- main changes or a new domain comes. In this work, we propose a data augmentation method with atomic templates for SLU, which involves minimum human efforts. The atomic templates produce exemplars for fine-grained constituents of semantic representations. We propose an encoder-decoder model to gener- ate the whole utterance from atomic exem- plars. Moreover, the generator could be trans- ferred from source domains to help a new do- main which has little data. Experimental re- sults show that our method achieves significant improvements on DSTC 2&3 dataset which is a domain adaptation setting of SLU. 1 Introduction The SLU module is a key component of spoken dialogue system, parsing user utterances into cor- responding semantic representations in a narrow domain. The typical semantic representation for SLU could be semantic frame (Tur and De Mori, 2011) or dialogue act (Young, 2007). In this pa- per, we focus on SLU with the dialogue act that a sentence is labelled as a set of act-slot-value triples. For example, the utterance “Not Chinese but I want Thai food please” has an annotation of “deny(food=Chinese), inform(food=Thai)”. Deep learning has achieved great success in the SLU field (Mesnil et al., 2015; Liu and Lane, 2016; Zhao et al., 2019). However, it is notori- ous for requiring large labelled data, which lim- its the scalability of SLU models. Despite re- ∗ Zijian Zhao and Su Zhu are co-first authors and con- tribute equally to this work. †The corresponding author is Kai Yu. Figure 1: Workflow of the data augmentation with atomic templates for SLU. cent advancements and tremendous research activ- ity in semi-supervised learning and domain adap- tation, the deep SLU models still require massive amounts of labelled data to train. Therefore, data augmentation for SLU becomes appealing, and it needs three kinds of capabilities: • Expression diversity: There are always vari- ous expressions for the same semantic mean- ing. Lack of expression diversity is an obsta- cle of SLU. • Semantic diversity: The data augmentation method should generate data samples with various semantic meanings. • Domain adaptation: The method should be able to utilize data from other domains and adapt itself to a new domain rapidly. In this work, we propose a new data augmen- tation method for SLU, which consists of atomic templates and a sentence generator (as shown in Figure 1). The method starts from dialogue acts 3638 which …

Content truncated to 500 words