Dialogue-Act Prediction of Future Responses Based on Conversation History

Abstract

Sequence-to-sequence models are a common approach to develop a chatbot. They can train a conversational model in an end-to-end manner. One significant drawback of such a neural network based approach is that the response generation process is a black-box, and how a specific response is generated is unclear. To tackle this problem, an interpretable response generation mechanism is desired. As a step toward this direction, we focus on dialogue-acts (DAs) that may provide insight to understand the response generation process. In particular, we propose a method to predict a DA of the next response based on the history of previous utterances and their DAs. Experiments using a Switch Board Dialogue Act corpus show that compared to the baseline considering only a single utterance, our model achieves 10.8% higher F1-score and 3.0% higher accuracy on DA prediction.

What Would a Teacher Do? Predicting Future Talk Moves

“…Figure 1. Following Tanaka et al (2019)'s model for future dialogue act prediction, its main components are three encoders. We hence name our model 3-E. Our model predicts the next teacher talk move t t+1 , given the last w context elements c t−w+1 , .…”

“…Dialogue act tagging, which is sometimes called dialogue act prediction, is the task of classifying an utterance into the category it belongs to (Yu and Yu, 2019;Khanpour et al, 2016;Wu et al, 2020). Analogous to FTMP, future dialogue act prediction is the task of predicting what the next dialogue act should be, given a conversation history (Tanaka et al, 2019).…”

Have You Made a Decision? Where? A Pilot Study on Interpretability of Polarity Analysis Based on Advising Problem

“…In order to verify our model's good performance we also import some strong models which take context in a dialogue into account. We choose RNN-based DA model [19], HRN [6], and ensemble HB which is proposed by us during DSTC8. For BERT-based model we select SUMBT [9].…”

Information Seeking in the Spirit of Learning: A Dataset for Conversational Curiosity

“…Inducing structure on conversations through dialog acts is helpful for analysis and downstream models (Tanaka et al, 2019). We introduce structurebeyond knowledge groundings-into Curiosity by annotating dialog acts for each message.…”

“…External Knowledge in Models Our model is related to those that incorporate external information like facts in question answering Sukhbaatar et al, 2015;Miller et al, 2016), knowledge base triples in dialog models (Han et al, 2015;He et al, 2017;Parthasarathi and Pineau, 2018), common sense (Young et al, 2018;Zhou et al, 2018a), or task-specific knowledge (Eric and Manning, 2017). Similarly to Kalchbrenner and Blunsom (2013); Khanpour et al (2016), CHARM predicts the act of the current message, but also next message's act like Tanaka et al (2019) do.…”