Learning to Select Context in a Hierarchical and Global Perspective for Open-domain Dialogue Generation
This work addresses context selection for open-domain dialogue generation, an incremental improvement over existing methods.
The paper tackles the challenge of selecting relevant context in open-domain multi-turn dialogue generation by addressing hierarchical semantic structure, redundant information, and long-term dependency. The proposed model with hierarchical self-attention and distant supervision outperforms baselines on two public datasets in fluency, coherence, and informativeness.
Open-domain multi-turn conversations mainly have three features, which are hierarchical semantic structure, redundant information, and long-term dependency. Grounded on these, selecting relevant context becomes a challenge step for multi-turn dialogue generation. However, existing methods cannot differentiate both useful words and utterances in long distances from a response. Besides, previous work just performs context selection based on a state in the decoder, which lacks a global guidance and could lead some focuses on irrelevant or unnecessary information. In this paper, we propose a novel model with hierarchical self-attention mechanism and distant supervision to not only detect relevant words and utterances in short and long distances, but also discern related information globally when decoding. Experimental results on two public datasets of both automatic and human evaluations show that our model significantly outperforms other baselines in terms of fluency, coherence, and informativeness.