Controlling hallucinations at word level in data-to-text generation

By Clément Rebuffel, Marco Roberti, Laure Soulier, Geoffrey Scoutheeten, Rossella Cancelliere, Patrick Gallinari. 2021

76 citations recorded by the source corpus.

Data-to-Text Generation (DTG) is a subfield of Natural Language Generation aiming at transcribing structured data in natural language descriptions. The field has been recently boosted by the use of neural-based generators which exhibit on one side great syntactic skills without the need of hand-crafted pipelines; on the other side, the quality of the generated text reflects the quality of the training data, which in realistic settings only offer imperfectly aligned structure-text pairs. Consequently, state-of-art neural models include misleading statements –usually called hallucinations—in their outputs. The control of this phenomenon is today a major challenge for DTG, and is the problem addressed in the paper. Previous work deal with this issue at the instance level: using an alignment score for each table-reference pair. In contrast, we propose a finer-grained approach, arguing that hallucinations should rather be treated at the word level. Specifically, we propose a Multi-Branch Decoder which is able to leverage word-level labels to learn the relevant parts of each training instance. These labels are obtained following a simple and efficient scoring procedure based on co-occurrence analysis and dependency parsing. Extensive evaluations, via automated metrics and human judgment on the standard WikiBio benchmark, show the accuracy of our alignment labels and the effectiveness of the proposed Multi-Branch Decoder. Our model is able to reduce and control hallucinations, while keeping fluency and coherence in generated texts. Further experiments on a degraded version of ToTTo show that our model could be successfully used on very noisy settings.

Subfields: Artificial Intelligence, Natural Language Processing, Generative Models

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

In Generative Models

Citations in the corpus

Survey of Hallucination in Natural Language Generation (2022, background): “, parts of the target sentence are loyal to the source input, while others diverge [110].”
Decoder-Only or Encoder-Decoder? Interpreting Language Model as a Regularized Encoder-Decoder (2023, methodology): “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.”
Factuality Enhanced Language Models for Open-Ended Text Generation (2022, background): “Hallucination in downstream NLG tasks There are active efforts to reduce the unfaithfulness or factual errors of task-speciﬁc LMs ﬁne-tuned for various downstream natural language generation (NLG) tasks such as summarization [42–48], data-to-text [49, 50, 20, 51–53] and dialogue system [54–58].”
A Token-level Reference-free Hallucination Detection Benchmark for Free-form Text Generation (2021, background): “Such reference-based hallucination detection has been proposed for ab-stractive summarization (Maynez et al., 2020), machine translation (Wang and Sennrich, 2020), data-to-text generation (Rebuffel et al., 2021), and image caption generation (Rohrbach et al., 2018).”
Siren's Song in the AI Ocean: A Survey on Hallucination in Large Language Models (2023, unknown): “Examples include the multi-branch decoder (Rebuffel et al., 2022) and the uncertainty-aware decoder (Xiao and Wang, 2021).”
Data-QuestEval: A Referenceless Metric for Data-to-Text Semantic Evaluation (2021, methodology): “8033 are more complex (see Table 3 in supplementary material); ii) the systems used for the evaluation of WikiBIO are very recent; their fluency is close to human level – see Table 4 in (Rebuffel et al., 2021).”