ikr7 / arxiv-survey

21 stars 0 forks source link

Is this Dialogue Coherent? Learning from Dialogue Acts and Entities #15661

Open arxiv-survey-bot[bot] opened 4 years ago

arxiv-survey-bot[bot] commented 4 years ago

URL: http://arxiv.org/abs/2006.10157v1

In this work, we investigate the human perception of coherence in open-domain dialogues. In particular, we address the problem of annotating and modeling the coherence of next-turn candidates while considering the entire history of the dialogue. First, we create the Switchboard Coherence (SWBD-Coh) corpus, a dataset of human-human spoken dialogues annotated with turn coherence ratings, where next-turn candidate utterances ratings are provided considering the full dialogue context. Our statistical analysis of the corpus indicates how turn coherence perception is affected by patterns of distribution of entities previously introduced and the Dialogue Acts used. Second, we experiment with different architectures to model entities, Dialogue Acts and their combination and evaluate their performance in predicting human coherence ratings on SWBD-Coh. We find that models combining both DA and entity information yield the best performances both for response selection and turn coherence rating.