Lune

ACL Findings2021

Probabilistic Graph Reasoning for Natural Proof Generation

Changzhi Sun, Xinbo Zhang, Jiangjie Chen, Chun Gan, Yuanbin Wu, Jiaze Chen, Hao Zhou, Lei Li

Abstract

In this paper, we investigate the problem of reasoning over natural language statements. Prior neural based approaches do not explicitly consider the inter-dependency among answers and their proofs. In this paper, we propose PROBR, a novel approach for joint answer prediction and proof generation. PROBR defines a joint probabilistic distribution over all possible proof graphs and answers via an induced graphical model. We then optimize the model using variational approximation on top of neural textual representation. Experiments on multiple datasets under diverse settings (fully supervised, few-shot and zero-shot evaluation) verify the effectiveness of PROBR, e.g., achieving 10%-30% improvement on QA accuracy in few/zero-shot evaluation. Our codes and models can be found at https://github.com/ changzhisun/PRobr/.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

4 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Learning Deductive Reasoning from Synthetic Corpus based on Formal Logic cites Generating Natural Language Proofs with Verifier-Guided Search
  • Generating Natural Language Proofs with Verifier-Guided Search cites PRover: Proof Generation for Interpretable Reasoning over Rules
  • ProofInfer: Generating Proof via Iterative Hierarchical Inference cites Generating Natural Language Proofs with Verifier-Guided Search
  • ProofInfer: Generating Proof via Iterative Hierarchical Inference cites PRover: Proof Generation for Interpretable Reasoning over Rules
  • PRover: Proof Generation for Interpretable Reasoning over Rules cites Learning to Reason: Leveraging Neural Networks for Approximate DNF Counting
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers4

  1. Learning Deductive Reasoning from Synthetic Corpus based on Formal LogicTerufumi Morishita, Gaku Morio, Atsuki Yamaguchi, Yasuhiro SogawaICML 2023 · 45 citations
  2. Generating Natural Language Proofs with Verifier-Guided SearchKaiyu Yang, Jia Deng, Danqi ChenEMNLP 2022 · 22 citations
  3. Logic-Regularized Verifier Elicits Reasoning from LLMsXinyu Wang, Changzhi Sun, Lian Cheng, Yuanbin Wu et al.ACL 2025 · 1 citation
  4. ProofInfer: Generating Proof via Iterative Hierarchical InferenceZichu Fei, Qi Zhang, Xin Zhou, Tao Gui et al.EMNLP 2022

Builds on2

  1. Learning to Reason: Leveraging Neural Networks for Approximate DNF CountingRalph Abboud, Ismail Ilkan Ceylan, Thomas LukasiewiczAAAI 2020 · 32 citations
  2. PRover: Proof Generation for Interpretable Reasoning over RulesSwarnadeep Saha, Sayan Ghosh, Shashank Srivastava, Mohit BansalEMNLP 2020 · 3 citations

Related papers

  1. Inductive Relation Prediction by BERTHanwen Zha, Zhiyu Chen, Xifeng YanAAAI 2022 · 69 citations
  2. ProofWriter: Generating Implications, Proofs, and Abductive Statements over Natural LanguageOyvind Tafjord, Bhavana Dalvi, Peter ClarkACL Findings 2021 · 122 citations
  3. Dynamic Neuro-Symbolic Knowledge Graph Construction for Zero-shot Commonsense Question AnsweringAntoine Bosselut, Ronan Le Bras, Yejin ChoiAAAI 2021 · 135 citations
  4. Coupled Variational Reinforcement Learning for Language Model General ReasoningXueru Wen, Jie Lou, Yanjiang Liu, Hongyu Lin et al.ICML 2026 · 1 citation
  5. Differentiable Reasoning on Large Knowledge Bases and Natural LanguagePasquale Minervini, Matko Bosnjak, Tim Rocktäschel, Sebastian Riedel et al.AAAI 2020 · 94 citations
Dusk over the sea between two cliffs drawn in fine vertical lines