Improving Dialogue State Tracking by Joint Slot Modeling
This work addresses a specific issue in task-oriented dialogue systems for improving accuracy, but it is incremental as it builds on existing models.
The paper tackled the problem of slot confusion in dialogue state tracking by proposing joint slot modeling methods, resulting in a state-of-the-art improvement from 58.7 to 61.3 on the MultiWoZ 2.1 dataset.
Dialogue state tracking models play an important role in a task-oriented dialogue system. However, most of them model the slot types conditionally independently given the input. We discover that it may cause the model to be confused by slot types that share the same data type. To mitigate this issue, we propose TripPy-MRF and TripPy-LSTM that models the slots jointly. Our results show that they are able to alleviate the confusion mentioned above, and they push the state-of-the-art on dataset MultiWoZ 2.1 from 58.7 to 61.3. Our implementation is available at https://github.com/CTinRay/Trippy-Joint.