SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 5478–5483 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 5478 SUMBT: Slot-Utterance Matching for Universal and Scalable Belief Tracking Hwaran Lee* Jinsik Lee* SK T-Brain, AI Center, SK telecom {hwaran.lee, jinsik16.lee, oceanos}@sktbrain.com Tae-Yoon Kim Abstract In goal-oriented dialog systems, belief track- ers estimate the probability distribution of slot- values at every dialog turn. Previous neu- ral approaches have modeled domain- and slot-dependent belief trackers, and have diffi- culty in adding new slot-values, resulting in lack of flexibility of domain ontology con- figurations. In this paper, we propose a new approach to universal and scalable be- lief tracker, called slot-utterance matching be- lief tracker (SUMBT). The model learns the relations between domain-slot-types and slot- values appearing in utterances through atten- tion mechanisms based on contextual seman- tic vectors. Furthermore, the model pre- dicts slot-value labels in a non-parametric way. From our experiments on two dialog cor- pora, WOZ 2.0 and MultiWOZ, the proposed model showed performance improvement in comparison with slot-dependent methods and achieved the state-of-the-art joint accuracy. 1 Introduction As the prevalent use of conversational agents, goal-oriented systems have received increasing at- tention from both academia and industry. The goal-oriented systems help users to achieve goals such as making restaurant reservations or booking flights at the end of dialogs. As the dialog pro- gresses, the system is required to update a distribu- tion over dialog states which consist of users’ in- tent, informable slots, and requestable slots. This is called belief tracking or dialog state tracking (DST). For instance, for a given domain and slot- types (e.g., ‘restaurant’ domain and ‘food’ slot- type), it estimates the probability of corresponding slot-value candidates (e.g., ‘Korean’ and ‘Modern *Hwaran Lee and Jinsik Lee equally contributed to this work. European’) that are pre-defined in a domain ontol- ogy. Since the system uses the predicted outputs of DST to choose the next action based on a dialog policy, the accuracy of DST is crucial to improve the overall performance of the system. Moreover, dialog systems should be able to deal with newly added domains and slots1 in a flexible manner, and thus developing scalable dialog state trackers is in- evitable. Regarding to this, Chen et al. (2016) has proposed a model to capture relations from intent- utterance pairs for intent expansion. Traditional statistical belief trackers (Hender- son et al., 2014) are vulnerable to lexical and morphological variations because they depend on manually constructed semantic dictionaries. With the rise of deep learning approaches, several neu- ral belief trackers (NBT) have been proposed and improved the performance by learning semantic neural representations of words (Mrkˇsi´c et al., 2017; Mrkˇsi´c and Vuli´c, 2018). However, the scalability still remains as a challenge; the previ- ously proposed methods either individually model each domain and/or slot (Zhong et al., 2018; Ren et al., 2018; Goel et al., 2018) or have difficulty in adding new slot-values that are not defined in the ontology (Ramadan …

Content truncated to 500 words