Learning to Identify Follow-Up Questions in Conversational Question Answering
Souvik Kundu, Qian Lin, Hwee Tou Ng
Question Answering Long Paper
Session 1B: Jul 6
(06:00-07:00 GMT)
Session 2A: Jul 6
(08:00-09:00 GMT)
Abstract:
Despite recent progress in conversational question answering, most prior work does not focus on follow-up questions. Practical conversational question answering systems often receive follow-up questions in an ongoing conversation, and it is crucial for a system to be able to determine whether a question is a follow-up question of the current conversation, for more effective answer finding subsequently. In this paper, we introduce a new follow-up question identification task. We propose a three-way attentive pooling network that determines the suitability of a follow-up question by capturing pair-wise interactions between the associated passage, the conversation history, and a candidate follow-up question. It enables the model to capture topic continuity and topic shift while scoring a particular candidate follow-up question. Experiments show that our proposed three-way attentive pooling network outperforms all baseline systems by significant margins.
You can open the
pre-recorded video
in a separate window.
NOTE: The SlidesLive video may display a random order of the authors.
The correct author list is shown at the top of this webpage.