Joint Modeling of Content and Discourse Relations in Dialogues

Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pages 974–984 Vancouver, Canada, July 30 - August 4, 2017. c⃝2017 Association for Computational Linguistics https://doi.org/10.18653/v1/P17-1090 Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pages 974–984 Vancouver, Canada, July 30 - August 4, 2017. c⃝2017 Association for Computational Linguistics https://doi.org/10.18653/v1/P17-1090 Joint Modeling of Content and Discourse Relations in Dialogues Kechen Qin1 Lu Wang1 Joseph Kim2 1College of Computer and Information Science, Northeastern University 2Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology 1qin.ke@husky.neu.edu, luwang@ccs.neu.edu 2joseph kim@csail.mit.edu Abstract We present a joint modeling approach to iden- tify salient discussion points in spoken meet- ings as well as to label the discourse rela- tions between speaker turns. A variation of our model is also discussed when discourse relations are treated as latent variables. Ex- perimental results on two popular meeting cor- pora show that our joint model can outperform state-of-the-art approaches for both phrase- based content selection and discourse rela- tion prediction tasks. We also evaluate our model on predicting the consistency among team members’ understanding of their group decisions. Classifiers trained with features constructed from our model achieve signif- icant better predictive performance than the state-of-the-art. 1 Introduction Goal-oriented dialogues, such as meetings, nego- tiations, or customer service transcripts, play an important role in our daily life. Automatically ex- tracting the critical points and important outcomes from dialogues would facilitate generating sum- maries for complicated conversations, understand- ing the decision-making process of meetings, or analyzing the effectiveness of collaborations. We are interested in a specific type of dia- logues — spoken meetings, which is a common way for collaboration and idea sharing. Previ- ous work (Kirschner et al., 2012) has shown that discourse structure can be used to capture the main discussion points and arguments put forward during problem-solving and decision-making pro- cesses in meetings. Indeed, content of different speaker turns do not occur in isolation, and should be interpreted within the context of discourse. Meanwhile, content can also reflect the purpose of speaker turns, thus facilitate with discourse rela- tion understanding. Take the meeting snippet from D: Three different types of batteries. Um can either use a hand dynamo, or the kinetic type ones, you know that they use in watches, or else uh a solar powered one. B: Um the bat uh the battery for a a watch wouldn't require a lot of power, would be my one query. Is a kinetic one going to be able to supply enough power? D: Yeah, I don't think it would. C: Yeah. D: We should probably just use conventional batteries. B: Which I suppose as well would allow us to go off the shelf again, you'd say ? D: Yeah. Uncertain Option Figure 1: A sample clip from AMI meeting corpus. B, C, and D denotes different speakers. Here we highlight salient phrases (in italics) that are relevant to the major topic discussed, i.e., “which type of battery to use for the remote control”. …

Content truncated to 500 words