We introduce a computational cognitive model for overlap resolution in articial agents. The model was inspired by findings from Conversation Analysis, and works by identifying the function of an overlap based on its position in the dialogue. Policies to drop out or maintain the turn are based on the type of overlap, the local discourse context, and the agent's goals. We integrate the model in the DIARC robotic architecture and show that it reproduces many of the behaviors found in the behavioral data.
@inproceedings{gervitsetal2018sigdial,
title={Pardon the Interruption: Managing Turn-Taking through Overlap Resolution in Embodied Artificial Agents},
author={Felix Gervits and Matthias Scheutz},
year={2018},
booktitle={Proceedings of the Special Interest Group on Discourse and Dialogue (Sigdial)},
url={https://hrilab.tufts.edu/publications/gervitsetal2018sigdial.pdf}
}