TY - GEN
T1 - Training Two-Stage Knowledge-Grounded Dialogues with Attention Feedback
AU - Li, Zhen
AU - Feng, Jiazhan
AU - Tao, Chongyang
AU - Zhao, Dongyan
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - Knowledge-grounded retrieval-based dialogue systems have attracted more and more attention. Among them, the two-stage dialogue models which separate the training stage into knowledge retrieving (via a retriever) and response ranking (via a ranker) are proved powerful. However, these approaches require knowledge-grounded dialogues with corresponding hand-annotated knowledge labels. Therefore, in this paper, we propose training two-stage knowledge-grounded dialogues with knowledge attention feedback from the ranker to the retriever. In each training iteration, the ranker provides knowledge attention scores as pseudo supervised feedback for the optimization of retriever. We conduct experiments on two public data sets. The experimental results demonstrate that our proposed method is superior to the existing baselines.
AB - Knowledge-grounded retrieval-based dialogue systems have attracted more and more attention. Among them, the two-stage dialogue models which separate the training stage into knowledge retrieving (via a retriever) and response ranking (via a ranker) are proved powerful. However, these approaches require knowledge-grounded dialogues with corresponding hand-annotated knowledge labels. Therefore, in this paper, we propose training two-stage knowledge-grounded dialogues with knowledge attention feedback from the ranker to the retriever. In each training iteration, the ranker provides knowledge attention scores as pseudo supervised feedback for the optimization of retriever. We conduct experiments on two public data sets. The experimental results demonstrate that our proposed method is superior to the existing baselines.
KW - Knowledge attention feedback
KW - Knowledge-grounded dialogues
KW - Multi-turn context modeling
KW - Retrieval-based dialogues
UR - https://www.scopus.com/pages/publications/85140441794
U2 - 10.1007/978-3-031-17120-8_37
DO - 10.1007/978-3-031-17120-8_37
M3 - 会议稿件
AN - SCOPUS:85140441794
SN - 9783031171192
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 473
EP - 484
BT - Natural Language Processing and Chinese Computing - 11th CCF International Conference, NLPCC 2022, Proceedings
A2 - Lu, Wei
A2 - Huang, Shujian
A2 - Hong, Yu
A2 - Zhou, Xiabing
PB - Springer Science and Business Media Deutschland GmbH
T2 - 11th CCF International Conference on Natural Language Processing and Chinese Computing, NLPCC 2022
Y2 - 24 September 2022 through 25 September 2022
ER -