Sticking to the Facts: Confident Decoding for Faithful Data-to-Text Generation

By Ran Tian, Thibault Sellam, Ankur P. Parikh. 2019

102 citations recorded by the source corpus.

Neural conditional text generation systems have achieved significant progress in recent years, showing the ability to produce highly fluent text. However, the inherent lack of controllability in these systems allows them to hallucinate factually incorrect phrases that are unfaithful to the source, making them often unsuitable for many real world systems that require high degrees of precision. In this work, we propose a novel confidence oriented decoder that assigns a confidence score to each target position. This score is learned in training using a variational Bayes objective, and can be leveraged at inference time using a calibration technique to promote more faithful generation. Experiments on a structured data-to-text dataset -- WikiBio -- show that our approach is more faithful to the source than existing state-of-the-art approaches, according to both automatic metrics and human evaluation.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Survey of Hallucination in Natural Language Generation (2022, background): “[134] propose a conidence score and a variational Bayes training framework to learn the score from data.”
BLEURT: Learning Robust Metrics for Text Generation (2020, background): “However, progress is increasingly impeded by the shortcomings of existing metrics (Wiseman et al., 2017; Ma et al., 2019; Tian et al., 2019).”
ToTTo: A Controlled Table-To-Text Generation Dataset (2020, background): “…hallucination , i.e., generating text that is ﬂuent but not faithful to the source (Vinyals and Le, 2015; Koehn and Knowles, 2017; Lee et al., 2018; Tian et al., 2019) and it is often easier to assess faithfulness of the generated text when the source content is structured (Wiseman et al., 2017;…”
Hurdles to Progress in Long-form Question Answering (2021, neutral): “A possible explanation for this issue is high source-reference divergence, a common problem in table-to-text generation (Wiseman et al., 2017; Tian et al., 2019).”
Get Your Vitamin C! Robust Fact Verification with Contrastive Evidence (2021, background): “…texts that match given facts is a known challenge (Fan et al., 2020; Kryscinski et al., 2020; Lewis et al., 2020b; Parikh et al., 2020; Shah et al., 2020; Tian et al., 2020) as language models tend to degenerate and hallucinate (Holtzman et al., 2020; Schuster et al., 2020; Zhou et al., 2020).”
Detecting Hallucinated Content in Conditional Neural Sequence Generation (2020, methodology): “…that standard metrics used for text evaluation, such as BLEU scores (Pa-pineni et al., 2002; Post, 2018), ROUGE (Lin and Hovy, 2004) and BERTScore (Zhang et al., 2019), do not correlate well with the faithfulness of model outputs (Maynez et al., 2020; Wang and Sennrich, 2020; Tian et al., 2019).”
Neural Path Hunter: Reducing Hallucination in Dialogue Systems via Path Grounding (2021, background): “C L ] 1 7 A pr 2 02 1 et al., 2020; Tian et al., 2019).”
Data-to-text Generation with Macro Planning (2021, unknown): “…are represented (Puduppully et al., 2019b; Iso et al., 2019), allowing the decoder to skip low-conﬁdence to-kens to enhance faithful generation (Tian et al., 2019), and making the encoder-decoder architecture more modular by introducing micro planning (Puduppully et al., 2019a; Moryossef et…”
Text Generation by Learning from Demonstrations (2020, unknown): “Recently, there is an increasing amount of work on discourag-12 ing implausible samples during training, e.g., using negative sampling (Welleck et al., 2020b), self-training on high-quality samples (Kedzie & McKeown, 2019), and conﬁdence-oriented decoding with calibration (Tian et al., 2020).”
Improved Natural Language Generation via Loss Truncation (2020, methodology): “For example, neural language models often produce ﬂuent text that is unfaithful to the source (Tian et al., 2019; Wiseman et al., 2017; Lee et al., 2018).”
Decoder-Only or Encoder-Decoder? Interpreting Language Model as a Regularized Encoder-Decoder (2023, background): “Generation models prone to generate unrelated content beyond the input data and this phenomenon has been recognized as the hallucination problem (Tian et al., 2019; Nie et al., 2019; Liu et al., 2021; Rebuffel et al., 2021) in many recent research works.”
Sometimes We Want Ungrammatical Translations (2021, background): “…input to an output that is not faithful to the source, and sometimes is even factually incorrect (Vinyals and Le, 2015; Koehn and Knowles, 2017; Wiseman et al., 2017; Nie et al., 2019; Kryscinski et al., 2020; Maynez et al., 2020; Tian et al., 2020; González et al., 2020; Xiao and Wang, 2021).”
Towards Faithfulness in Open Domain Table-to-text Generation from an Entity-centric View (2021, methodology): “For faithfulness evaluation, prior work introduces soft constraints (Tian et al. 2019; Wang et al. 2020b), template-like (Liu et al. 2017; Wiseman, Shieber, and Rush 2018; Shen et al. 2020; Li and Rush 2020; Ye et al. 2020) or controllable generation (Peng et al. 2018; Fan, Lewis, and Dauphin 2019;…”
Towards Faithful Neural Table-to-Text Generation with Content-Matching Constraints (2020, methodology): “Speciﬁcally, Tian et al. (2019) develop a conﬁdence oriented decoder that assigns a conﬁdence score to each target position to reduce the unfaithful information in the generated text.”
AgreeSum: Agreement-Oriented Multi-Document Summarization (2021, background): “More broadly, there have been other recent works striving to develop techniques for high-precision generation (Malmi et al., 2019; Tian et al., 2019; Pang and He, 2020; Parikh et al., 2020; Dušek and Kasner, 2020).”
A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions (2023, unknown): “…such as abstractive summarization (Kryscinski et al., 2020; Wang et al., 2020a; Maynez et al., 2020; Fabbri et al., 2021; Goyal and Durrett, 2021; Pagnoni et al., 2021; Tang et al., 2022), data-to-text(Tian et al., 2019; Parikh et al., 2020), and machine translation (Zhou et al., 2021).”
Focus Attention: Promoting Faithfulness and Diversity in Summarization (2021, background): “Tian et al. (2019) learn a confidence score to ensure that the model attends to the source whenever necessary.”