Pith. sign in

REVIEW 5 major objections 5 minor 45 references

The paper argues that context-free causal direction questions are best answered by explicitly constructing a target-aware causal graph, auditing each candidate chain counterfactually, and aggregating signed path evidence.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 15:41 UTC pith:IL23RXQG

load-bearing objection Novel, well-specified framework for auditable LLM causal reasoning; empirical claims need tighter evaluation before they can be believed. the 5 major comments →

arxiv 2607.15281 v1 pith:IL23RXQG submitted 2026-04-22 cs.AI

Causal-Audit: Explicit and Auditable Graph-based Reasoning via Target-Aware Causal Chain Construction

Su Lan , Xuefei Yin , Yanming Zhu , Alan Wee-Chung Liew This is my paper
classification cs.AI
keywords causal reasoningintervention-based question answeringtarget-aware causal graphcounterfactual auditingevidence aggregationlarge language modelscontext-free QAinterpretability
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tries to show that context-free causal questions—"if I intervene on X, does Y become more or less likely?"—are answered more accurately by building and auditing an explicit causal graph than by letting an LLM reason implicitly. Its framework identifies the intervention and target, expands a directed graph outward from X while keeping the base target as a hard constraint, extracts candidate causal chains, audits each edge by counterfactual probing, and aggregates the signed evidence into a decision. The reported result is consistent accuracy gains over direct prompting and free-form deliberation baselines across three benchmarks and several small instruction-tuned backbones, with the reasoning trace exposed for inspection. The paper notes that the outputs are supportive reasoning signals, not formal do-calculus causal identification.

Core claim

On its own terms, the paper's discovery is that a context-free LLM can be converted from an end-to-end predictor into a constrained causal evaluator: after extracting a structured query (X, surface target, base target, modifiers), the model is asked to propose single-hop causal triples, but expansion is pruned by fine-grained alignment to the base target (entity, quantity, and state), and only paths that survive premise consistency and per-edge counterfactual verification are aggregated. The aggregation treats each path as a signed product of edge polarities, weights paths by audit score and a bridge-edge penalty, and forms a graph confidence from the normalized difference between positive a

What carries the argument

The load-bearing object is a target-aware causal graph: a directed graph expanded by breadth-first search from the intervention X, where the base target Yb is used both as a relevance classifier (EXACT, CLOSEHIT, BRIDGE, NONE via entity, quantity, and state alignment) and as a ranking constraint for top-K pruning. On top of it, path-level causal evidence aggregation computes each path's net sign as the product of edge polarities, weights it by a counterfactual audit score and a bridge-edge penalty, and combines positive and negative evidence mass into graph confidence Cgraph = |W+ − W−|/(W+ + W− + ε) · log(1 + |P∗|), which captures both evidence dominance and support size. These mechanisms d

Load-bearing premise

The load-bearing premise is that the benchmark labels the method is measured against are genuine causal effects; on the medical benchmark they are association statistics p(Y|X) − p(Y|¬X), which the paper itself calls data-driven directional proxies rather than fully identified causal effects, so if association and intervention diverge under confounding the accuracy numbers may track correlation rather than causation.

What would settle it

Construct a synthetic causal model with known true graph and a confounder such that the association between X and Y is opposite in sign to the true intervention effect P(Y|do(X)) − P(Y|do(¬X)); render it in the paper's context-free question format, run the framework, and see whether it answers according to the do-effect or the association proxy. Matching the proxy on disagreements would falsify the claim that the framework performs causal rather than correlational reasoning.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the claim holds, context-free intervention question answering no longer has to be end-to-end: any instruction-tuned LLM can be turned into a structured reasoner whose intermediate causal hypotheses can be inspected and traced.
  • Treating the target as a core constraint during graph expansion suppresses irrelevant variables, which the results suggest improves consistency across model families and domains, especially where free-form deliberation baselines are high-variance.
  • Path-level auditing combined with signed aggregation outperforms single-chain reasoning and also outperforms a supervised fine-tuning approach that learns causal DAGs, without requiring task-specific fine-tuning.
  • Bounded depth with moderate branching and longer path lengths is the best operating regime; deeper and wider expansion degrades accuracy, directly motivating the pruning and auditing design.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: because the medical benchmark labels are association-based proxies (p(Y|X) − p(Y|¬X)), the reported gains may partly reflect better prediction of association, not true interventional effect; a decisive test requires benchmarks with known do-calculus ground truth.
  • Editorial inference: the same LLM proposes edges and then verifies them, creating a possible self-confirmation risk; a stronger audit would use a different model or a symbolic knowledge base for verification.
  • Editorial inference: the aggregation formula's confidence conflates evidence dominance with support size; a natural extension is calibrated abstention or uncertainty estimates when positive and negative evidence are nearly balanced.
  • Editorial inference: the four-stage decomposition suggests a testable diagnostic—systematically swap surface modifiers and negations to check whether the mapping back to surface semantics is stable.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. This paper introduces Causal-Audit, a framework for context-free intervention-based question answering. It extracts causal variables from the question, constructs a target-aware causal graph via bounded BFS expansion with LLM-proposed causal triples, extracts and audits causal chains via premise and counterfactual checks, and aggregates signed path evidence with an LLM fallback to predict whether an intervention makes a target more or less likely. Experiments on DDXPlus-CausalEffect, a WIQA directional subset, and a CauseNet-derived dataset report accuracy gains over Direct, CoT, ToT, GoT, and CDCR-SFT baselines across three LLM backbones. The paper also includes hyperparameter sensitivity analysis and a case study.

Significance. The proposed framework is a reasonable and timely contribution: it makes the reasoning process explicit and auditable, introduces target-aware pruning to suppress irrelevant variables, and aggregates counterfactually scored causal paths. The pipeline is internally coherent, and the sensitivity analysis in Table 3 and the case study in Appendix D add useful qualitative insight. If properly validated, the approach could be a valuable inference-time alternative to fine-tuning for causal direction QA. However, the empirical evidence as presented does not yet support the paper's central causal claim because two of the three benchmarks rest on association-based proxy labels and the reporting lacks uncertainty quantification. The strengths are the transparent staged design, the auditable graph traces, and the effort to move beyond end-to-end prediction.

major comments (5)
  1. [Table 1 / Appendix B.2 (Eqs. 12-13)] The DDXPlus-CausalEffect gold labels are computed from the association contrast Δ = p(Y|X) − p(Y|¬X) and thresholded in Eq. (13). The paper itself calls these 'data-driven directional proxies rather than fully identified causal effects' and the Limitations state that the framework 'does not perform formal causal identification based on do-calculus.' Under confounding or reverse causation, the sign of Δ can disagree with the sign of the interventional effect of do(X) on Y. Reporting accuracy against these labels in Table 1 therefore measures association-direction prediction, not causal-effect prediction, and directly weakens the abstract's claim of 'causal reasoning traces.' To support the causal claim, the paper should add an evaluation with interventional ground truth (e.g., the WIQA subset alone, or a simulated do-calculus benchmark) or a confounding-adjusted analysis.
  2. [Table 1 caption] The caption states 'Accuracy is reported under a best-of-multiple-attempts setting,' but the protocol is not defined anywhere in the paper. It is unclear how many attempts are made, what selection criterion is used, and whether the same protocol is applied to all baselines. If only the proposed method uses best-of-k, the comparison is biased. Please specify the protocol precisely and apply it uniformly, or report single-attempt results.
  3. [Tables 1 and 4] The test sets are small (N=200, 212, 100), and no confidence intervals, bootstrap estimates, or significance tests are reported. Several reported gaps are small; for example, on WIQA Llama-3.1-8B, 67.92 vs 66.04 (Table 1) is a 1.88-point difference on 212 examples. The claim that the framework 'consistently outperforms' all baselines requires uncertainty quantification before it can be accepted.
  4. [Appendix C.4 (P2, P4, P5)] The same LLM proposes causal triples (P2), validates them counterfactually (P4), and renders the final decision (P5). The 'audit' is therefore a self-consistency check by the same model, not an independent verification of causal validity. This does not by itself invalidate the benchmark comparisons, but it limits the auditable/verifiability claim. Please quantify the agreement between P2 and P4, or audit with a different model, or temper the wording.
  5. [Section 4.2 vs Table 3] It is unclear whether the hyperparameters (D=4, R=3, L=6) were selected on a held-out development split or by inspecting test accuracy in Table 3. If the latter, the main results are optimistically biased. Please clarify the split used for hyperparameter selection and, ideally, report the sensitivity analysis on a development set rather than on the test set.
minor comments (5)
  1. [References] Besta et al. (2024a) and (2024b) are the same paper; Wei et al. (2022a) and (2022b) are also the same. Duplicate entries should be merged.
  2. [Appendix B.2] In Eq. (12), the quantities n_{XY}, n_X, n_Y, and N are not defined. Please add explicit definitions.
  3. [Table 4] The column 'Neg. (%)' is not described in the main text. Please clarify how negation is detected and whether it affects the label mapping.
  4. [Figure 1] The example in Figure 1 discusses Guillain-Barré syndrome, while the case study in Section 4.6 concerns Chagas. Consider labeling the figure as a separate illustration or aligning it with the case study to avoid confusion.
  5. [Reproducibility] No code or data artifact link is provided. Given the multi-stage LLM pipeline and benchmark construction details, releasing code and the constructed datasets would substantially aid reproducibility.

Circularity Check

0 steps flagged

No significant circularity: the central derivation is self-contained and benchmarked on external or separately constructed labels.

full rationale

I walked the claimed derivation chain: structured query extraction (Eq. 1), target-aware BFS graph construction (Eqs. 2-4), chain extraction and counterfactual auditing (Eqs. 5-6), and evidence aggregation (Eqs. 7-11). None of these equations is fit to the gold labels, and no fitted parameter is renamed as a prediction. The final decision rule (Eq. 11) combines graph evidence with an LLM conditioned on that evidence; it is not a tautology. Evaluation uses WIQA labels from the original benchmark, DDXPlus-CausalEffect labels computed from association statistics (Eqs. 12-13), and CauseNet labels from path polarity products (Eq. 14). These labels are used only for evaluation, not as inputs to the method. The paper itself flags in Appendix B.2 that the DDXPlus labels are 'data-driven directional proxies rather than fully identified causal effects' and in Limitations that the framework 'does not perform formal causal identification based on do-calculus.' That is a validity limitation of the causal claim, not a circularity in the derivation. The same LLM proposes and audits edges (P2/P4), which weakens the independence of the audit, but this is not a by-construction equivalence between the predicted answer and the input. There are no load-bearing self-citations or imported uniqueness theorems. Therefore the circularity score is 0.

Axiom & Free-Parameter Ledger

10 free parameters · 6 axioms · 0 invented entities

The central claim rests mainly on the reliability of LLM-generated causal triples and on the validity of benchmark labels. The DDXPlus labels are acknowledged proxies, the CauseNet labels inherit extractor noise, and several aggregation constants are unreported. Free parameters are mostly set on dev split; γ, τ_or, τ, ε, and K' are unspecified.

free parameters (10)
  • Expansion depth D = D=4 (chosen on dev split)
    Controls how many hops the graph grows; sensitivity analysis (Table 3) shows performance varies with D, so central results depend on this choice.
  • Frontier width K (Top-K) = K=2
    Only the top-2 ranked nodes are kept each BFS step (Eq. 4); no ablation for K is reported.
  • Branching width R = R=3
    Number of candidate triples generated per frontier node; sensitivity Table 3 shows an interaction with path length.
  • Max path length L = L=6
    Paths longer than L are excluded; sensitivity analysis shows performance varies with L.
  • Audit threshold τ_audit = 0.6
    Paths with multiplicative audit score below τ_audit are discarded (Eq. 6).
  • Bridge-edge decay γ = not reported
    Appears in path weight Eq. (8) as exp(−γ N_br); no value is given anywhere in the paper.
  • Decision threshold τ_or = not reported
    Graph decision is trusted only when C_graph > τ_or (Eq. 11); threshold is not specified.
  • DDXPlus label margin τ = not reported
    Instances with |Δ| ≤ τ are discarded when building DDXPlus-CausalEffect labels (Eq. 13); affects benchmark composition.
  • Evidence epsilon ε = not reported
    Small constant in Eq. (9) to avoid division by zero; not numerically specified.
  • Number of audited paths K' = not reported
    Top-K' paths are retained for audit and final evidence; no value is given.
axioms (6)
  • domain assumption LLM-proposed causal triples (u,r,v) are plausible and causally directed.
    The graph is built entirely from Φ_expand (Eq. 2); any systematic LLM bias enters the graph.
  • domain assumption Counterfactual edge scores V_cf(e) from the same LLM are reliable estimates of true causal validity.
    Eq. (6) multiplies these scores, and Appendix C.4 shows the same model both proposes and verifies edges; there is no external check.
  • domain assumption Association statistics on DDXPlus approximate interventional effects.
    Appendix B.2 labels are p(Y|X)−p(Y|¬X); the authors call them proxies rather than causal effects.
  • domain assumption CauseNet edge signs and polarity products define ground-truth causal direction.
    Appendix B.4 generates labels from sign products of sampled paths, inheriting open-IE extraction noise.
  • domain assumption Path-effect sign product (Eq. 7) gives net causal effect without interactions or confounders.
    Assumes monotonic composition along chains; no interaction terms are modeled.
  • ad hoc to paper Evidence aggregation C_graph (Eq. 9) is a valid confidence measure.
    Hand-designed combination of evidence dominance and path count; no derivation or calibration is provided.

pith-pipeline@v1.3.0-alltime-deepseek · 14640 in / 15111 out tokens · 155380 ms · 2026-08-02T15:41:01.363441+00:00 · methodology

0 comments
read the original abstract

Causal and intervention-based question answering is fundamental to advancing large language models (LLMs) toward reasoning beyond surface-level correlations and understanding underlying causal mechanisms. However, existing LLM-based methods often rely on implicit language-level reasoning, resulting in opaque causal assumptions, unverifiable reasoning paths, and fragile predictions under complex interventions, particularly in context-free settings. In this paper, we propose an explicit and auditable causal reasoning framework for context-free intervention-based question answering. Our method formulates causal inference as structured reasoning over an explicit causal graph through four modular stages, rather than implicit end-to-end prediction. A key innovation is a target-aware causal graph construction strategy that treats the target variable as a core constraint during graph expansion, effectively suppressing irrelevant variables, spurious causal relations, and reasoning noise. We further introduce a path-level causal evidence aggregation mechanism that combines multiple causal paths while modeling both reinforcing and counteracting effects, enabling robust decision-making beyond single-chain reasoning. Extensive experiments on three benchmarks demonstrate that our framework consistently outperforms existing LLM-based methods while providing interpretable and auditable causal reasoning traces.

Figures

Figures reproduced from arXiv: 2607.15281 by Alan Wee-Chung Liew, Su Lan, Xuefei Yin, Yanming Zhu.

Figure 1
Figure 1. Figure 1: Overview of the proposed explicit causal reasoning framework for context-free intervention-based question [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 8 linked inside Pith

  1. [1]

    Aho and Jeffrey D

    Alfred V. Aho and Jeffrey D. Ullman , title =. 1972

    1972

  2. [2]

    Publications Manual , year = "1983", publisher =

    1983

  3. [3]

    Chandra and Dexter C

    Ashok K. Chandra and Dexter C. Kozen and Larry J. Stockmeyer , year = "1981", title =. doi:10.1145/322234.322243

    arXiv 1981

  4. [4]

    Scalable training of

    Andrew, Galen and Gao, Jianfeng , booktitle=. Scalable training of

  5. [5]

    Dan Gusfield , title =. 1997

    1997

  6. [6]

    Tetreault , title =

    Mohammad Sadegh Rasooli and Joel R. Tetreault , title =. Computing Research Repository , volume =. 2015 , url =

    2015

  7. [7]

    , author=

    A framework for learning predictive structures from multiple tasks and unlabeled data. , author=. Journal of machine learning research , volume=

  8. [8]

    2020 , publisher =

    Causal Inference: What If , author =. 2020 , publisher =

    2020

  9. [9]

    Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    Mitigating hallucinations in large language models via causal reasoning , author=. Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    2025

  10. [10]

    Advances in neural information processing systems , volume=

    Ddxplus: A new dataset for automatic medical diagnosis , author=. Advances in neural information processing systems , volume=

  11. [11]

    WIQA : A dataset for ``What if...'' reasoning over procedural text

    Tandon, Niket and Dalvi, Bhavana and Sakaguchi, Keisuke and Clark, Peter and Bosselut, Antoine. WIQA : A dataset for ``What if...'' reasoning over procedural text. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019

    2019

  12. [12]

    Proceedings of the 29th ACM international conference on information & knowledge management , pages=

    Causenet: Towards a causality graph extracted from the web , author=. Proceedings of the 29th ACM international conference on information & knowledge management , pages=

  13. [13]

    First Workshop on Foundations of Reasoning in Language Models , year=

    CARE: Turning LLMs Into Causal Reasoning Expert , author=. First Workshop on Foundations of Reasoning in Language Models , year=

  14. [14]

    Thirty-Ninth Annual Conference on Neural Information Processing Systems , year=

    Counterfactual reasoning: an analysis of in-context emergence , author=. Thirty-Ninth Annual Conference on Neural Information Processing Systems , year=

  15. [15]

    Proceedings of the 31st International Conference on Computational Linguistics , pages=

    Towards Faithful Multi-step Reasoning through Fine-Grained Causal-aware Attribution Reasoning Distillation , author=. Proceedings of the 31st International Conference on Computational Linguistics , pages=

  16. [16]

    arXiv preprint arXiv:2501.14892 , year=

    Causal graphs meet thoughts: Enhancing complex reasoning in graph-augmented llms , author=. arXiv preprint arXiv:2501.14892 , year=

    Pith/arXiv arXiv

  17. [17]

    Advances in Neural Information Processing Systems , volume=

    Cladder: Assessing causal reasoning in language models , author=. Advances in Neural Information Processing Systems , volume=

  18. [18]

    Proceedings of the 29th International Conference on Computational Linguistics , pages=

    CausalQA: A benchmark for causal question answering , author=. Proceedings of the 29th International Conference on Computational Linguistics , pages=

  19. [19]

    C ausal B ench: A Comprehensive Benchmark for Evaluating Causal Reasoning Capabilities of Large Language Models

    Wang, Zeyu. C ausal B ench: A Comprehensive Benchmark for Evaluating Causal Reasoning Capabilities of Large Language Models. Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing. 2024

    2024

  20. [20]

    Advances in Neural Information Processing Systems , volume=

    Unveiling causal reasoning in large language models: Reality or mirage? , author=. Advances in Neural Information Processing Systems , volume=

  21. [21]

    arXiv preprint arXiv:2404.06349 , year=

    Causalbench: A comprehensive benchmark for causal learning capability of llms , author=. arXiv preprint arXiv:2404.06349 , year=

    Pith/arXiv arXiv

  22. [22]

    Com ^2 : A Causal-Guided Benchmark for Exploring Complex Commonsense Reasoning in Large Language Models

    Xiong, Kai and Ding, Xiao and Cao, Yixin and Yan, Yuxiong and Du, Li and Zhang, Yufei and Gao, Jinglong and Liu, Jiaqian and Qin, Bing and Liu, Ting. Com ^2 : A Causal-Guided Benchmark for Exploring Complex Commonsense Reasoning in Large Language Models. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Lon...

    2025

  23. [23]

    Advances in neural information processing systems , volume=

    Chain-of-thought prompting elicits reasoning in large language models , author=. Advances in neural information processing systems , volume=

  24. [24]

    Advances in neural information processing systems , volume=

    Large language models are zero-shot reasoners , author=. Advances in neural information processing systems , volume=

  25. [25]

    The Eleventh International Conference on Learning Representations , year=

    Self-Consistency Improves Chain of Thought Reasoning in Language Models , author=. The Eleventh International Conference on Learning Representations , year=

  26. [26]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Graph of thoughts: Solving elaborate problems with large language models , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  27. [27]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Large language models are neurosymbolic reasoners , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  28. [28]

    Proceedings of the 32nd ACM International Conference on Multimedia , pages=

    Causal-driven large language models with faithful reasoning for knowledge question answering , author=. Proceedings of the 32nd ACM International Conference on Multimedia , pages=

  29. [29]

    Causality: objectives and assessment , pages=

    Causal inference , author=. Causality: objectives and assessment , pages=. 2010 , publisher=

    2010

  30. [30]

    Making Reasoning Matter: Measuring and Improving Faithfulness of Chain-of-Thought Reasoning

    Paul, Debjit and West, Robert and Bosselut, Antoine and Faltings, Boi. Making Reasoning Matter: Measuring and Improving Faithfulness of Chain-of-Thought Reasoning. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024

    2024

  31. [31]

    Findings of the Association for Computational Linguistics ACL 2024 , pages=

    Knowledge Graph-Enhanced Large Language Models via Path Selection , author=. Findings of the Association for Computational Linguistics ACL 2024 , pages=

    2024

  32. [32]

    Causal Reasoning in Large Language Models: A Knowledge Graph Approach , author=. 2024

    2024

  33. [33]

    Proceedings of the ACM on Web Conference 2025 , pages=

    Paths-over-graph: Knowledge graph empowered large language model reasoning , author=. Proceedings of the ACM on Web Conference 2025 , pages=

    2025

  34. [34]

    arXiv preprint arXiv:2504.07087 , year=

    KG-LLM-Bench: A Scalable Benchmark for Evaluating LLM Reasoning on Textualized Knowledge Graphs , author=. arXiv preprint arXiv:2504.07087 , year=

    Pith/arXiv arXiv

  35. [35]

    2005 , publisher=

    Making things happen: A theory of causal explanation , author=. 2005 , publisher=

    2005

  36. [36]

    2009 , edition =

    Causality: Models, Reasoning, and Inference , author =. 2009 , edition =

    2009

  37. [37]

    arXiv preprint arXiv:2508.21010 , year=

    ChainReaction! Structured Approach with Causal Chains as Intermediate Representations for Improved and Explainable Causal Video Question Answering , author=. arXiv preprint arXiv:2508.21010 , year=

    Pith/arXiv arXiv

  38. [38]

    arXiv preprint arXiv:2307.07697 , year=

    Think-on-graph: Deep and responsible reasoning of large language model on knowledge graph , author=. arXiv preprint arXiv:2307.07697 , year=

    Pith/arXiv arXiv

  39. [39]

    2022 , issn =

    Causality-aware Enhanced Model for Multi-hop Question Answering over Knowledge Graphs , journal =. 2022 , issn =

    2022

  40. [40]

    arXiv preprint arXiv:2305.08291 , year=

    Large language model guided tree-of-thought , author=. arXiv preprint arXiv:2305.08291 , year=

    Pith/arXiv arXiv

  41. [41]

    Advances in neural information processing systems , volume=

    Tree of thoughts: Deliberate problem solving with large language models , author=. Advances in neural information processing systems , volume=

  42. [42]

    arXiv preprint arXiv:2410.15319 , year=

    Causality for large language models , author=. arXiv preprint arXiv:2410.15319 , year=

    Pith/arXiv arXiv

  43. [43]

    2018 , publisher =

    The Book of Why: The New Science of Cause and Effect , author =. 2018 , publisher =

    2018

  44. [44]

    arXiv preprint arXiv:2410.06392 , year=

    Counterfactual causal inference in natural language with large language models , author=. arXiv preprint arXiv:2410.06392 , year=

    Pith/arXiv arXiv

  45. [45]

    The design and analysis of algorithms , pages=

    Depth-first and breadth-first search , author=. The design and analysis of algorithms , pages=. 1992 , publisher=

    1992