Cross-domain Semantic Parsing via Paraphrasing

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 1235–1246 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Cross-domain Semantic Parsing via Paraphrasing Yu Su Department of Computer Science University of California, Santa Barbara ysu@cs.ucsb.edu Xifeng Yan Department of Computer Science University of California, Santa Barbara xyan@cs.ucsb.edu Abstract Existing studies on semantic parsing mainly focus on the in-domain setting. We formulate cross-domain semantic parsing as a domain adaptation problem: train a semantic parser on some source domains and then adapt it to the target domain. Due to the diversity of logical forms in different domains, this problem presents unique and intriguing challenges. By con- verting logical forms into canonical utter- ances in natural language, we reduce se- mantic parsing to paraphrasing, and de- velop an attentive sequence-to-sequence paraphrase model that is general and flex- ible to adapt to different domains. We discover two problems, small micro vari- ance and large macro variance, of pre- trained word embeddings that hinder their direct use in neural networks, and pro- pose standardization techniques as a rem- edy. On the popular OVERNIGHT dataset, which contains eight domains, we show that both cross-domain training and stan- dardized pre-trained word embedding can bring significant improvement. 1 Introduction Semantic parsing, which maps natural language utterances into computer-understandable logical forms, has drawn substantial attention recently as a promising direction for developing natural lan- guage interfaces to computers. Semantic pars- ing has been applied in many domains, includ- ing querying data/knowledge bases (Woods, 1973; Zelle and Mooney, 1996; Berant et al., 2013), con- trolling IoT devices (Campagna et al., 2017), and communicating with robots (Chen and Mooney, 2011; Tellex et al., 2011; Artzi and Zettlemoyer, 2013; Bisk et al., 2016). Despite the wide applications, studies on se- mantic parsing have mainly focused on the in- domain setting, where both training and testing data are drawn from the same domain. How to build semantic parsers that can learn across do- mains remains an under-addressed problem. In this work, we study cross-domain semantic pars- ing. We model it as a domain adaptation prob- lem (Daum´e III and Marcu, 2006), where we are given some source domains and a target domain, and the core task is to adapt a semantic parser trained on the source domains to the target domain (Figure 1). The benefits are two-fold: (1) by train- ing on the source domains, the cost of collecting training data for the target domain can be reduced, and (2) the data of source domains may provide in- formation complementary to the data collected for the target domain, leading to better performance on the target domain. This is a very challenging task. Traditional domain adaptation (Daum´e III and Marcu, 2006; Blitzer et al., 2006) only concerns natural lan- guages, while semantic parsing concerns both nat- ural and formal languages. Different domains of- ten involve different predicates. In Figure 1, from the source BASKETBALL domain a semantic parser can learn the semantic mapping from natural lan- guage to …

Content truncated to 500 words