Open-Domain Dialogue Quality Evaluation: Deriving Nugget-level Scores from Turn-level Scores
This work addresses the need for more precise evaluation to improve dialogue systems, though it appears incremental as it builds on existing turn-level methods.
The paper tackles the problem of fine-grained dialogue quality evaluation by decomposing system turns into nuggets and deriving nugget-level scores from existing turn-level systems, demonstrating potential effectiveness through a case study.
Existing dialogue quality evaluation systems can return a score for a given system turn from a particular viewpoint, e.g., engagingness. However, to improve dialogue systems by locating exactly where in a system turn potential problems lie, a more fine-grained evaluation may be necessary. We therefore propose an evaluation approach where a turn is decomposed into nuggets (i.e., expressions associated with a dialogue act), and nugget-level evaluation is enabled by leveraging an existing turn-level evaluation system. We demonstrate the potential effectiveness of our evaluation method through a case study.