SCIENCE@home

Context-aware multi-feature fusion for open-domain dialogue generation

Jing Liao,Hong Yu,Qi Cheng,Ye Wang

Published 2022 in International Conference on Industrial Technology

ABSTRACT

With the rapid development of deep learning, open-domain dialogue models are able to generate relatively fluent conversations. However, current generative models present incomplete representations of dialogue features, which make it prone to generate low-quality or irrelevant-semantic responses. To address this problem, this paper presents a novel context-aware multi-feature fusion model, integrating keywords and category-semantic information into the dialogue generation. Specifically, distinct keywords share various significance in a sentence, incorporating this significance into the conversation text can improve the generative ability of the generative model. Our method enhances the feature representation of the text and thus generates high-quality conversation. Finally, we conducted comprehensive experiments on the DailyDialog dataset and EmpatheticDialogues dataset, analyzing the experimental results and verifying the feasibility of our approach.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-28 of 28 references · Page 1 of 1

CITED BY