EMNLP Findings2022
Controllable Factuality in Document-Grounded Dialog Systems Using a Noisy Channel Model
Nico Daheim, David Thulke, Christian Dugast, Hermann Ney
摘要
In this work, we present a model for documentgrounded response generation in dialog that is decomposed into two components according to Bayes' theorem. One component is a traditional ungrounded response generation model and the other component models the reconstruction of the grounding document based on the dialog context and generated response. We propose different approximate decoding schemes and evaluate our approach on multiple opendomain and task-oriented document-grounded dialog datasets. Our experiments show that the model is more factual in terms of automatic factuality metrics than the baseline model. Furthermore, we outline how introducing scaling factors between the components allows for controlling the tradeoff between factuality and fluency in the model output. Finally, we compare our approach to a recently proposed method to control factuality in grounded dialog, CTRL (Rashkin et al., 2021), and show that both approaches can be combined to achieve additional improvements.
问 Lune
几分钟读透这篇 Paper,不用再花一下午。
Lune 能读完全文,也能读引用它的顶会 Paper,讲清它的结论、证据和局限,每一点都注明出自哪篇 Paper。不用硬啃,也不靠猜。
脉络
- Retrieval Augmentation Reduces Hallucination in Conversation 引用了 BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension
- Retrieval Augmentation Reduces Hallucination in Conversation 引用了 Poly-encoders: Architectures and Pre-training Strategies for Fast and Accurate Multi-sentence Scoring
- Retrieval Augmentation Reduces Hallucination in Conversation 引用了 Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
- Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks 引用了 BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension
它引用的 Paper13
- Retrieval-Augmented Generation for Knowledge-Intensive NLP TasksPatrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni 等NeurIPS 2020 · 19,162 次引用
- BERTScore: Evaluating Text Generation with BERTTianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger 等ICLR 2020 · 8,443 次引用
- BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and ComprehensionMike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad 等ACL 2020 · 1,224 次引用
- Retrieval Augmentation Reduces Hallucination in ConversationKurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela 等EMNLP Findings 2021 · 579 次引用
- Poly-encoders: Architectures and Pre-training Strategies for Fast and Accurate Multi-sentence ScoringSamuel Humeau, Kurt Shuster, Marie-Anne Lachaux, Jason WestonICLR 2020 · 316 次引用
相关 Paper
- Increasing Faithfulness in Knowledge-Grounded Dialogue with Controllable FeaturesHannah Rashkin, David Reitter, Gaurav Singh Tomar, Dipanjan DasACL 2021
- Pointwise Mutual Information Based Metric and Decoding Strategy for Faithful Generation in Document Grounded DialogsYatin Nandwani, Vineet Kumar, Dinesh Raghu, Sachindra Joshi 等EMNLP 2023 · 4 次引用
- Enhancing Factuality through Consensus and Consistency in Summarization Using Minimum Bayes Risk DecodingRiza Setiawan Soetedjo, Yusuke Sakai, Hidetaka Kamigaito, Jingun Kwon 等ACL Findings 2026
- A Controllable Model of Grounded Response GenerationZeqiu Wu, Michel Galley, Chris Brockett, Yizhe Zhang 等AAAI 2021 · 89 次引用
- RetGen: A Joint Framework for Retrieval and Grounded Text Generation ModelingYizhe Zhang, Siqi Sun, Xiang Gao, Yuwei Fang 等AAAI 2022 · 45 次引用
