Template Guided Text Generation for Task-Oriented Dialogue

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 6505–6520, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 6505 Template Guided Text Generation for Task-Oriented Dialogue Mihir Kale Google Mountain View mihirkale@google.com Abhinav Rastogi Google Research Mountain View abhirast@google.com Abstract Virtual assistants such as Google Assistant, Amazon Alexa, and Apple Siri enable users to interact with a large number of services and APIs on the web using natural language. In this work, we investigate two methods for Nat- ural Language Generation (NLG) using a sin- gle domain-independent model across a large number of APIs. First, we propose a schema- guided approach which conditions the genera- tion on a schema describing the API in natu- ral language. Our second method investigates the use of a small number of templates, grow- ing linearly in number of slots, to convey the semantics of the API. To generate utterances for an arbitrary slot combination, a few sim- ple templates are first concatenated to give a semantically correct, but possibly incoherent and ungrammatical utterance. A pre-trained language model is subsequently employed to rewrite it into coherent, natural sounding text. Through automatic metrics and human evalua- tion, we show that our method improves over strong baselines, is robust to out-of-domain in- puts and shows improved sample efficiency. 1 1 Introduction Virtual assistants have become popular in recent years and task-completion is one of their most im- portant aspects. These assistants help users in accomplishing tasks such as finding restaurants, buying sports tickets, finding the weather etc., by providing a natural language interface to many ser- vices or APIs available on the web. Most systems include a natural language understanding and dia- logue state tracking module for semantic parsing of the dialogue history. This is followed by a policy module which interacts with the APIs, whenever re- quired, and generates the actions to be taken by the 1Our code and data is available at github.com/google- research/schema-guided-dialogue system to continue the dialog. In the end, the Natu- ral Language Generation (NLG) module converts these actions into an utterance, which is surfaced to the user. Being the user-facing interface of the dialogue system, NLG is one of the most important components impacting user experience. Traditional NLG systems heavily utilize a set of templates to produce system utterances. Al- though the use of templates gives good control over the outputs generated by the system, defining templates becomes increasingly tedious as more APIs are added. Supporting multi-domain conver- sations spanning multiple APIs quickly grows out of hand, requiring expert linguists and rigorous testing to ensure the grammatical correctness and appropriateness of generated utterances. Conse- quently, data-driven generative approaches have gained prominence. Such systems require much less effort and can generate utterances containing novel patterns. Meanwhile, with the rapid prolifer- ation of personal assistants, supporting large num- ber of APIs across multiple domains has become increasingly important, resulting in research on supporting new APIs with few labelled examples (few-shot learning). To this end, generative models pre-trained on large …

Content truncated to 500 words