Can NLI Models Verify QA Systems' Predictions?

By Jifan Chen, Eunsol Choi, Greg Durrett. 2021

65 citations recorded by the source corpus.

To build robust question answering systems, we need the ability to verify whether answers to questions are truly correct, not just"good enough"in the context of imperfect QA datasets. We explore the use of natural language inference (NLI) as a way to achieve this goal, as NLI inherently requires the premise (document context) to contain all necessary information to support the hypothesis (proposed answer to the question). We leverage large pre-trained models and recent prior datasets to construct powerful question converter and decontextualization modules, which can reformulate QA instances as premise-hypothesis pairs with very high reliability. Then, by combining standard NLI datasets with NLI examples automatically derived from QA training data, we can train NLI models to judge the correctness of QA models' proposed answers. We show that our NLI approach can generally improve the confidence estimation of a QA model across different domains, evaluated in a selective QA setting. Careful manual analysis over the predictions of our NLI model shows that it can further identify cases where the QA model produces the right answer for the wrong reason, or where the answer cannot be verified as addressing all aspects of the question.

Subfields: Artificial Intelligence, Information Systems, 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 Information Systems

In Natural Language Processing

Citations in the corpus

WANLI: Worker and AI Collaboration for Natural Language Inference Dataset Creation (2022, background): “, 2018), and can be applied to verify candidate answers in question-answering (Chen et al., 2021) or factuality of generated summaries (Maynez et al.”
REV: Information-Theoretic Evaluation of Free-Text Rationales (2022, background): “For NLI, the vacuous baseline rationale was obtained after paraphrasing. model fine-tuned on a set of ( question , answer , declarative sentence ) tuples (Demszky et al., 2018) following Chen et al. (2021b).”
The Unreliability of Explanations in Few-shot Prompting for Textual Reasoning (2022, methodology): “Following past work on supervised calibration (Kamath et al., 2020; Chen et al., 2021; Ye and Durrett, 2022), we can learn a calibrator that tunes the probabilities of a prediction based on the score of its explanation (Section 4.2).”
Impossible Distillation for Paraphrasing and Summarization: How to Make High-quality Lemonade out of Small, Low-quality Model (2023, background): “NLI models are well-suited to quantify this relationship, as they are trained to infer the logical entailment between an arbitrary pair of statements (Chen et al., 2021).”
The Unreliability of Explanations in Few-Shot In-Context Learning (2022, methodology): “Following past work on supervised calibration (Kamath et al., 2020; Chen et al., 2021; Ye and Durrett, 2022), we can learn a calibrator that tunes the probabilities of a prediction based on the score of its explanation (Section 4.2).”