An Empirical Study of Topic Transition in Dialogue

Transitioning between various topics is a natural component of human-human. Although, topic transitions has been studied in dialogue for decades, only a handful of corpora based studies have been performed to analyze the subtleties in topic transitions. Thus, this study annotates $215$ conversations from switchboard corpus and studies how factors such as length of a conversation, number of topic transitions, topic transitions by participants are related. This paper presents an empirical study on topic transition in switchboard corpus followed by modelling topic transition with a precision of $83\%$ for in-domain($id$) test set and $82\%$ on $10$ \textit{out-of-domain} ($ood$). It is envisioned that this work will be fundamental in emulating human-human like topic transition in open-domain dialog systems.