We aim to build a data-driven virtual assistant or a chat companion system with the aid of big data and deep learning techniques.
A Survey on Dialogue Systems: Recent Advances and New Frontiers
Open Data Sets
We release a public available JD customer service corpus, consisting of online retailing customer service dialogues. In JD corpus, each conversation is between a customer and a customer service staff. The corpus contains more than 420,000 dialogues sessions.
To access the corpus, please visit here.
Hierarchical Variational Memory Network for Dialogue Generation [cite]
We propose a novel hierarchical variational memory network (HVMN), by adding the hierarchical structure and the variational memory network into a neural encoder-decoder network. By emulating human-to-human dialogues, our proposed method can capture both the high-level abstract variations and long-term memories during dialogue tracking, which enables the random access of relevant dialogue histories.
Knowledge Diffusion for Neural Dialogue Generation
End-to-end neural dialogue generation has shown promising results recently, but it does not employ knowledge to guide the generation and hence tends to generate short, general, and meaningless responses. We propose a neural knowledge diffusion (NKD) model to introduce knowledge into dialogue generation. This method can not only match the relevant facts for the input utterance but diffuse them to similar entities. With the help of facts matching and entity diffusion, the neural dialogue generation is augmented with the ability of convergent and divergent thinking over the knowledge base.