Research map / Large language models / Agents, code and math
From Document Layout to Causal Topology: An End-to-End Architecture for Temporal Causal Reasoning
Zhifeng Yang, Wei Luo, Yongjian Chen, Shiqi Ou
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes reframing temporal causal reasoning from discrete narratives to causal topologies, with a training-free neuro-symbolic agent (TCR-Agent) and a second architecture, based only on the truncated abstract.
Problem
LLMs generally lack deep causal reasoning in temporal reasoning tasks.
Approach
Inspired by document layout reordering, shifts from discrete event narratives to causal topologies; a training-free neuro-symbolic agent (TCR-Agent) performs interpretable causal deduction and counterfactual truncation.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add DAG instantiation to TCR-Agent (ablation) | TCR-Agent (GPT-5.4) · TempoCausal-Bench | LLM-Acc: 63.1 -> 70.4 | Improved |
| add dual-track reasoning to TCR-Agent (ablation) | TCR-Agent (GPT-5.4) · TempoCausal-Bench | LLM-Acc: 65.2 -> 70.4 | Improved |
| add counterfactual intervention to TCR-Agent (ablation) | TCR-Agent (GPT-5.4) · TempoCausal-Bench | LLM-Acc: 67.5 -> 70.4 | Improved |
| add scratchpad to TCR-Agent (ablation) | TCR-Agent (GPT-5.4) · TempoCausal-Bench | LLM-Acc: 66.8 -> 70.4 | Improved |
| add causal-topological attention mask to TCR-Former (ablation) | TCR-Former (Qwen-3-8B) · TempoCausal-Bench | LLM-Acc: 67.0 -> 75.7 | Improved |
| add fact-logic bypass fusion to TCR-Former decoder (ablation) | TCR-Former (Qwen-3-8B) · TempoCausal-Bench | LLM-Acc: 71.4 -> 75.7 | Improved |
| add time span bias to TCR-Former (ablation) | TCR-Former (Qwen-3-8B) · TempoCausal-Bench | LLM-Acc: 72.3 -> 75.7 | Improved |
| add a causal-topological encoder with a Qwen-3-0.6B backbone without the proposed mask and biases (Only Encoder + Decoder) versus decoder alone (ablation) | TCR-Former (Qwen-3-8B) · TempoCausal-Bench | LLM-Acc: 58.6 -> 55.4 | Hurt |
| replace external prompt methods (TG-LLM, Time-CoT, TISER) with TCR-Agent | GPT-5.4 · TempoCausal-Bench | LLM-Acc: 59.2 -> 70.4 | Improved |
| internalize causal topology into the model (TCR-Former) rather than LoRA fine-tuning on TCR-Agent outputs | Qwen-3-8B · TempoCausal-Bench | LLM-Acc: 67.4 -> 75.7 | Improved |
| add TCR-Agent external graph and CoT reasoning (versus TISER) for rule inference errors | Qwen-3-8B · TempoCausal-Bench error attribution | Rule inference mismatch error proportion: 13.8 (TISER) -> 13.0 (TCR-Agent) | No clear effect |
| use TCR-Former instead of TCR-Agent for inference cost | Qwen-3-8B · TempoCausal-Bench | inference latency: -62% latency, about -44% output tokens | Improved |
Limitations
- Only the abstract was available; full method and results are unknown.
Closest papers by meaning
- Mitigating Hallucinations in Large Language Models via Causal Reasoning (AAAI 2026 · similarity 0.89)
- Unveiling Causal Reasoning in Large Language Models: Reality or Mirage? (NeurIPS 2024 · similarity 0.89)
- Language Agents Meet Causality -- Bridging LLMs and Causal World Models (ICLR 2025 · similarity 0.88)
- NeSTR: A Neuro-Symbolic Abductive Framework for Temporal Reasoning in Large Language Models (AAAI 2026 · similarity 0.88)
- COLD: Causal reasOning in cLosed Daily activities (NeurIPS 2024 · similarity 0.87)
- Nuance Matters: Probing Epistemic Consistency in Causal Reasoning (AAAI 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.
