FaiRR: Faithful and Robust Deductive Reasoning over Natural Language

TL;DR

FaiRR introduces a modular, faithful, and robust transformer-based framework for deductive reasoning over natural language, ensuring transparent inference steps and improved robustness compared to black-box models.

Contribution

The paper proposes a modular reasoning framework with independent components to enhance faithfulness and robustness in natural language deductive reasoning.

Findings

FaiRR is more robust to language perturbations.

FaiRR achieves faster inference times.

Errors are more interpretable due to modular design.

Abstract

Transformers have been shown to be able to perform deductive reasoning on a logical rulebase containing rules and statements written in natural language. Recent works show that such models can also produce the reasoning steps (i.e., the proof graph) that emulate the model's logical reasoning process. Currently, these black-box models generate both the proof graph and intermediate inferences within the same model and thus may be unfaithful. In this work, we frame the deductive logical reasoning task by defining three modular components: rule selection, fact selection, and knowledge composition. The rule and fact selection steps select the candidate rule and facts to be used and then the knowledge composition combines them to generate new inferences. This ensures model faithfulness by assured causal relation from the proof step to the inference reasoning. To test our framework, we propose FaiRR (Faithful and Robust Reasoner) where the above three components are independently modeled by transformers. We observe that FaiRR is robust to novel language perturbations, and is faster at inference than previous works on existing reasoning datasets. Additionally, in contrast to black-box generative models, the errors made by FaiRR are more interpretable due to the modular approach.