Retrieval-guided Dialogue Response Generation via a Matching-to-Generation Framework
Генерация ответов в диалоге с использованием поиска руководящих примеров через фреймворк «сопоставление‑к‑генерации»
2019-01-01
SCID: 54.1/qpnxv3vd
Discuss with AI
dialogue response generationmatching-to-generation frameworkretrieval-augmented generationretrieval-guided dialogue
Figures from the paper
Abstract (AI)
Deng Cai, Yan Wang, Wei Bi, Zhaopeng Tu, Xiaojiang Liu, Shuming Shi. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.
Key Findings
1
Approach reported in EMNLP-IJCNLP 2019 as a novel way to combine retrieval and generation for dialogue systems
2
Framework improves retrieval-guided dialogue response generation by matching retrieved candidates before generation
3
Method leverages retrieved responses as explicit guidance rather than directly copying them into generated replies
4
Proposes a matching-to-generation framework that uses retrieved dialogue responses to guide generation of new responses
Research Object
Retrieval-guided dialogue response generation system (matching-to-generation framework)
Research Subject
Effectiveness of a matching-to-generation framework for guiding neural dialogue response generation using retrieved candidate responses
Publication Details
Publication Date
2019-01-01
Journal
Publisher
ISSN
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest