A Weakly Supervised Learning Approach for Spoken Language Understanding

Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 2006), pages 199–207, Sydney, July 2006. c⃝2006 Association for Computational Linguistics A Weakly Supervised Learning Approach for Spoken Language Understanding Wei-Lin Wu, Ru-Zhan Lu, Jian-Yong Duan, Hui Liu, Feng Gao, Yu-Quan Chen Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai, 200030, P. R. China {wu-wl,lu-rz,duan-jy,liuhui,gaofeng,chen-yq} @cs.sjtu.edu.cn Abstract In this paper, we present a weakly super- vised learning approach for spoken lan- guage understanding in domain-specific dialogue systems. We model the task of spoken language understanding as a suc- cessive classification problem. The first classifier (topic classifier) is used to iden- tify the topic of an input utterance. With the restriction of the recognized target topic, the second classifier (semantic classifier) is trained to extract the corre- sponding slot-value pairs. It is mainly data-driven and requires only minimally annotated corpus for training whilst re- taining the understanding robustness and deepness for spoken language. Most im- portantly, it allows the employment of weakly supervised strategies for training the two classifiers. We first apply the training strategy of combining active learning and self-training (Tur et al., 2005) for topic classifier. Also, we pro- pose a practical method for bootstrapping the topic-dependent semantic classifiers from a small amount of labeled sentences. Experiments have been conducted in the context of Chinese public transportation information inquiry domain. The experi- mental results demonstrate the effective- ness of our proposed SLU framework and show the possibility to reduce human labeling efforts significantly. 1 Introduction Spoken Language Understanding (SLU) is one of the key components in spoken dialogue systems. Its task is to identify the user’s goal and extract from the input utterance the information needed to complete the query. Traditionally, there are mainly two mainstreams in the SLU researches: knowledge-based approaches, which are based on robust parsing or template matching tech- niques (Sneff, 1992; Dowding et al., 1993; Ward and Issar, 1994); and data-driven approaches, which are generally based on stochastic models (Pieraccini and Levin, 1993; Miller et al., 1995). Both approaches have their drawbacks, however. The former approach is cost-expensive to de- velop since its grammar development is time- consuming, laboursome and requires linguistic skills. It is also strictly domain-dependent and hence difficult to be adapted to new domains. On the other hand, although addressing such draw- backs associated with knowledge-based ap- proaches, the latter approach often suffers the data sparseness problem and hence needs a fully annotated corpus in order to reliably estimate an accurate model. More recently, some new varia- tion methods are proposed through certain trade- offs, such as the semi-automatically grammar learning approach (Wang and Acero, 2001) and Hidden Vector State (HVS) model (He and Young, 2005). The two methods require only minimally annotated data (only the semantic frames are annotated). This paper proposes a novel weakly super- vised spoken language understanding approach. Our SLU framework mainly includes two suc- cessive classifiers: topic classifier and semantic classifier. The main advantage of the proposed approach is that it …

Content truncated to 500 words