Research map / Graph learning / Knowledge and temporal graphs
Discourse-Aware Graph Networks for Textual Logical Reasoning
Yinya Huang, Lemao Liu, Kun Xu, Meng Fang, Liang Lin, Xiaodan Liang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| initialize nodes from contextual EDU embeddings instead of random embeddings (ablation) | DAGNs · ReClor Test | accuracy (%): 56.40 -> 61.00 | Improved |
| add homogeneous variable edges within context and candidate nodes (ablation) | DAGNs · ReClor Test | accuracy (%): 61.00 -> 59.30 | Hurt |
| use discourse-based edge linking instead of fully-connected linking (ablation) | DAGNs · ReClor Test | accuracy (%): 56.10 -> 61.00 | Improved |
| use discourse-based edge linking instead of random linking (ablation) | DAGNs · ReClor Test | accuracy (%): 55.90 -> 61.00 | Improved |
| distinguish explicit and implicit connective edge types instead of a single type (ablation) | DAGNs · ReClor Test | accuracy (%): 57.70 -> 61.00 | Improved |
| use EDU nodes instead of clause nodes (ablation) | DAGNs · ReClor Test | accuracy (%): 56.60 -> 61.00 | Improved |
| use EDU nodes instead of sentence nodes (ablation) | DAGNs · ReClor Test | accuracy (%): 57.30 -> 61.00 | Improved |
| add logic edge reasoning (ablation) | DAGNs · ReClor Test | accuracy (%): 59.80 -> 61.00 | Improved |
| add global graph representation (ablation) | DAGNs · ReClor Test | accuracy (%): 60.10 -> 61.00 | Improved |
| add binary node types (ablation) | DAGNs without edge reasoning and global representation · ReClor Test | accuracy (%): 58.70 -> 59.50 | Improved |
| add graph reasoning with hierarchical fusion (ablation) | DAGNs · ReClor Test | accuracy (%): 52.00 -> 61.00 | Improved |
| add DAGNs logic graphs to RoBERTa-Large | RoBERTa-Large-MC · ReClor Test | accuracy (%): 55.60 -> 61.00 | Improved |
Closest papers by meaning
- Debate on Graph: A Flexible and Reliable Reasoning Framework for Large Language Models (AAAI 2025 · similarity 0.88)
- Graph Reasoning Transformers for Knowledge-Aware Question Answering (AAAI 2024 · similarity 0.87)
- Improving Complex Reasoning over Knowledge Graph with Logic-Aware Curriculum Tuning (AAAI 2025 · similarity 0.87)
- Argument Graph Uncertainty: Quantifying Uncertainty from the Logical Structure of Reasoning Chains (NeurIPS 2026 · similarity 0.87)
- Graph-constrained Reasoning: Faithful Reasoning on Knowledge Graphs with Large Language Models (ICML 2025 · similarity 0.87)
- Reasoning of Large Language Models over Knowledge Graphs with Super-Relations (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
