Multi-hop question answering requires composing multiple pieces of evidence to derive the correct answer, making it particularly vulnerable to hallucinated intermediate reasoning steps. A common failure mode arises when neural generators produce intermediate sentences that are fluent but logically inconsistent with the underlying facts, which can subsequently mislead downstream reasoning. In this paper, we propose ITERLOGIC, an iterative natural logic verification framework for multi-hop question answering, which explicitly integrates symbolic verification into the generation process. Specifically, a neural generator composes candidate intermediate sentences from retrieved facts, while a natural logic verifier evaluates their logical consistency with the source evidence. Low-scoring candidates are iteratively rejected and regenerated, allowing the system to progressively eliminate hallucinated compositions. This lightweight interaction upgrades natural logic from a passive scoring component to an active verification signal that guides intermediate sentence generation. Experiments on QASC, WorldTree, and WikiHop demonstrate that ITERLOGIC improves answer accuracy, highlighting the effectiveness of iterative verification for reliable multi-hop reasoning.

Entailment trees are essential for enhancing interpretability and transparency in tasks like question answering and natural language understanding. However, existing approaches often lack logical consistency, as they rely on static reward structures or ignore the intricate dependencies within multi-step reasoning. To address these limitations, we propose a method that integrates natural logic principles into reinforcement learning, enabling dynamic reward computation to guide entailment tree generation. Our approach ensures logical consistency across reasoning steps while improving interpretability and generalization. Experiments on EntailmentBank demonstrate significant improvements over state-of-the-art methods, highlighting the effectiveness of natural logic in structured reasoning.

Many open-domain question answering problems can be cast as a textual entailment task, where a question and candidate answers are concatenated to form hypotheses. A QA system then determines if the supporting knowledge bases, regarded as potential premises, entail the hypotheses. In this paper, we investigate a neural-symbolic QA approach that integrates natural logic reasoning within deep learning architectures, towards developing effective and yet explainable question answering models. The proposed model gradually bridges a hypothesis and candidate premises following natural logic inference steps to build proof paths. Entailment scores between the acquired intermediate hypotheses and candidate premises are measured to determine if a premise entails the hypothesis. As the natural logic reasoning process forms a tree-like, hierarchical structure, we embed hypotheses and premises in a Hyperbolic space rather than Euclidean space to acquire more precise representations. Empirically, our method outperforms prior work on answering multiple-choice science questions, achieving the best results on two publicly available datasets. The natural logic inference process inherently provides evidence to help explain the prediction process.