Extractive Summarization and Dialogue Act Modeling on Email Threads: An Integrated Probabilistic Approach
Tatsuro Oya, Giuseppe Carenini · 2014
In this paper, we present a novel supervised approach to the problem of summarizing email conversations and modeling dialogue acts. We assume that there is a relationship between dialogue acts and important sen-tences. Based on this assumption, we intro-duce a sequential graphical model approach which simultaneously summarizes email conversation and models dialogue acts. We compare our model with sequential and non-sequential models, which independent-ly conduct the tasks of extractive summari-zation and dialogue act modeling. An empirical evaluation shows that our ap-proach significantly outperforms all base-lines in classifying correct summary sentences without losing performance on dialogue act modeling task. 1