Chat, Shift and Perform: Bridging the Gap between Task-oriented and Non-task-oriented Dialog Systems
This work addresses the gap between casual and goal-driven dialog systems, offering a hybrid approach that is incremental in combining existing dialog types.
The authors tackled the problem of integrating open-domain chat and task-oriented dialog by proposing CASPER, a system with specialized models for chatting, topic shifting, and task performance, which in a user study improved naturalness, satisfaction, and task-elicitation rates compared to baselines.
We propose CASPER (ChAt, Shift and PERform), a novel dialog system consisting of three types of dialog models: chatter, shifter, and performer. Shifter, which is designed for topic switching, enables a seamless flow of dialog from open-domain chat- to task-oriented dialog. In a user study, CASPER gave a better impression in terms of naturalness of response, lack of forced topic switching, and satisfaction compared with a baseline dialog system trained in an end-to-end manner. In an ablation study, we found that naturalness of response, dialog satisfaction, and task-elicitation rate improved compared with when shifter was removed from CASPER, indicating that topic shift with shifter supports the introduction of natural task-oriented dialog.