Scaling Multi-Domain Dialogue State Tracking via Query Reformulation

Abstract

We present a novel approach to dialogue state tracking and referring expression resolution tasks. Successful contextual understanding of multi-turn spoken dialogues requires resolving referring expressions across turns and tracking the entities relevant to the conversation across turns. Tracking conversational state is particularly challenging in a multi-domain scenario when there exist multiple spoken language understanding (SLU) sub-systems, and each SLU sub-system operates on its domainspecific meaning representation. While previous approaches have addressed the disparate schema issue by learning candidate transformations of the meaning representation, in this paper, we instead model the reference resolution as a dialogue context-aware user query reformulation task -the dialog state is serialized to a sequence of natural language tokens representing the conversation. We develop our model for query reformulation using a pointer-generator network and a novel multi-task learning setup.In our experiments, we show a significant improvement in absolute F1 on an internal as well as a, soon to be released, public benchmark respectively. *

Can You Unpack That? Learning to Rewrite Questions-in-Context

“…An interesting avenue for future work is to incorporate deeper context, either from other modalities (Das et al, 2017) or from other dialog comprehension tasks . Parallel to our work, Rastogi et al (2019) and Su et al (2019) introduce utterance rewriting datasets for dialog state tracking. Rastogi et al (2019) covers a narrow set of domains and the rewrites of Su et al (2019) are based on Chinese dialog with two-turn fixed histories.…”

Learning to Generate Reformulation Actions for Scalable Conversational Query Understanding

“…Note that high BLEU score does not guarantee high accuracy. NLG metrics like BLEU is common for this kind of reformulation [7,8], where valid queries have significant overlaps with ground truth (more concise and less diverse than translations or answers). However, BLEU is still insufficient since improper reformulations may also resemble ground truth.…”

“…Conversation query understanding (CQU) is crucial to multi-turn question answering (QA), but remains challenging owing to diversity of queries (see Table 1). For Conversational AI, much work has been proposed for language understanding (LU) [6] and dialogue state tracking [7] in task-oriented dialogue system (DS). However, in multi-turn QA, user intents and conversational phenomena are more vague and complex (e.g., clarifications, coreferences and ellipsis), but there is no general LU solution as in DS pipelines.…”

“…Therefore, CQR is an independent module naturally adapting to downstream technologies. Instead of reformulating queries end-to-end [7,8,11], intermediate actions are designed to capture syntactic regularities of conversations, and new queries are synthesized from the actions. The regularities are imposed across domains by natural language itself (domain-independent), and the actions essentially decouple them from domain-specific entity information.…”

“What Do You Mean by That?” A Parser-Independent Interactive Approach for Enhancing Text-to-SQL

“…Another topic our method related to is query reformulation. The idea of query reformulation is explored by Ray et al (2018) and Rastogi et al (2019), while they apply this idea in other domains with different scenarios. Our work is also related to semantic parsing, the process of converting natural language utterances into logical forms.…”