Most of the current man-machine dialogues are at the two end-points of a spectrum of dialogues, i.e. goal-driven dialogues and non goal-driven chit-chats. Document-driven dialogues provide a bridge between them with the change of documents from structured data to unstructured free texts. This paper proposes a Document Driven Dialogue Generation model (D3G) which generates dialogues centering a given document, as well as answering user’s questions. A Doc-Reader mechanism is designed to locate the content related to user’s questions in documents. A Multi-Copy mechanism is employed to generate document-related responses. And the dialogue history is used in both mechanisms. Experimental results on the CMU_DOG dataset show that our D3G model can not only generate informative responses that are more relevant to the document, but also answer user’s questions better than the baseline models.
CITATION STYLE
Li, K., Bai, Z., Wang, X., & Yuan, C. (2019). A Document Driven Dialogue Generation Model. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11856 LNAI, pp. 508–520). Springer. https://doi.org/10.1007/978-3-030-32381-3_41
Mendeley helps you to discover research relevant for your work.