Global-to-local Memory Pointer Networks for Task-Oriented Dialogue
This addresses the problem of scalable knowledge integration in dialogue systems for developers and users, representing an incremental improvement with a novel method for a known bottleneck.
The paper tackles the challenge of incorporating large, dynamic knowledge bases into end-to-end task-oriented dialogue systems by proposing GLMP networks, which improve copy accuracy and mitigate out-of-vocabulary issues, achieving state-of-the-art results on bAbI Dialogue and Stanford Multi-domain Dialogue datasets.
End-to-end task-oriented dialogue is challenging since knowledge bases are usually large, dynamic and hard to incorporate into a learning framework. We propose the global-to-local memory pointer (GLMP) networks to address this issue. In our model, a global memory encoder and a local memory decoder are proposed to share external knowledge. The encoder encodes dialogue history, modifies global contextual representation, and generates a global memory pointer. The decoder first generates a sketch response with unfilled slots. Next, it passes the global memory pointer to filter the external knowledge for relevant information, then instantiates the slots via the local memory pointers. We empirically show that our model can improve copy accuracy and mitigate the common out-of-vocabulary problem. As a result, GLMP is able to improve over the previous state-of-the-art models in both simulated bAbI Dialogue dataset and human-human Stanford Multi-domain Dialogue dataset on automatic and human evaluation.