EMNLP Findings2024
SARCAT: Generative Span-Act Guided Response Generation using Copy-enhanced Target Augmentation
Jeong-Doo Lee, Hyeongjun Choi, Beomseok Hong, Youngsub Han, Byoung-Ki Jeon, Seung-Hoon Na
Abstract
In this paper, we present a novel extension to improve document-grounded response generation by proposing the Generative Span-Act Guided Response Generation Using Copy-Enhanced Target Augmentation (SARCAT), which consists of two major components: 1) Copy-enhanced target-side input augmentation is an extended data augmentation that handles the exposure bias problem by additionally incorporating the copy mechanism on top of target-side augmentation (Xie et al., 2021); 2) Span-act guided response generation first predicts the grounding spans and dialogue acts before generating a response. Experimental results on the validation set in MultiDoc2Dial show that the proposed SARCAT yields improvements over strong baselines in both seen and unseen settings, achieving start-of-the-art performance even with the base reader using the pretrained T5-base model.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Retrieval Augmentation Reduces Hallucination in Conversation cites Knowledge-Grounded Dialogue Generation with Pre-trained Language Models
- Retrieval Augmentation Reduces Hallucination in Conversation cites Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Builds on9
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni et al.NeurIPS 2020 · 19,162 citations
- Retrieval Augmentation Reduces Hallucination in ConversationKurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela et al.EMNLP Findings 2021 · 579 citations
- Knowledge-Grounded Dialogue Generation with Pre-trained Language ModelsXueliang Zhao, Wei Wu, Can Xu, Chongyang Tao et al.EMNLP 2020 · 153 citations
- MultiDoc2Dial: Modeling Dialogues Grounded in Multiple DocumentsSong Feng, Siva Sankalp Patel, Hui Wan, Sachindra JoshiEMNLP 2021 · 42 citations
Related papers
- Topic-Aware Response Generation in Task-Oriented Dialogue with Unstructured Knowledge AccessYue Feng, Gerasimos Lampouras, Ignacio IacobacciEMNLP Findings 2022 · 6 citations
- Causal Document-Grounded Dialogue Pre-trainingYingxiu Zhao, Bowen Yu, Bowen Li, Haiyang Yu et al.EMNLP 2023 · 2 citations
- KPT: Keyword-Guided Pre-training for Grounded Dialog GenerationQi Zhu, Fei Mi, Zheng Zhang, Yasheng Wang et al.AAAI 2023 · 5 citations
- PLATO: Pre-trained Dialogue Generation Model with Discrete Latent VariableSiqi Bao, Huang He, Fan Wang, Hua Wu et al.ACL 2020 · 229 citations
- Controllable Factuality in Document-Grounded Dialog Systems Using a Noisy Channel ModelNico Daheim, David Thulke, Christian Dugast, Hermann NeyEMNLP Findings 2022 · 1 citation
