relational causal identification

This paper addresses the identifiability of causal effects under causal abstraction—specifically, how to infer treatment effects when the underlying causal graph is incompletely known, as commonly encountered in high-dimensional or complex systems. Method: We propose the first hierarchical identifiability criterion system tailored to causal abstraction, systematically linking abstract causal structures at varying granularities to their corresponding identification capabilities. Our theoretical framework integrates causal graph models, observational data, and formal logical reasoning, yielding decidable, layered identifiability criteria. Contribution/Results: The framework does not require a fully specified causal graph; it enables identifiability assessment even when the causal structure is unknown or only partially known. We validate its effectiveness and practicality through rigorous analysis of canonical examples from the causal inference literature.

This work addresses the fundamental identifiability problem in causal inference—determining whether a target causal quantity is identifiable from available data. We propose a unified qualitative causal identification framework supporting three classes of queries: interventional, counterfactual, and cross-domain transportability. Methodologically, we model latent-variable causal structures using acyclic directed mixed graphs (ADMGs), integrate state-of-the-art identification algorithms, and design a domain-specific language (DSL) for declarative specification and symbolic reasoning over causal expressions. Our contributions are threefold: (1) the first unification of diverse causal query types under a single identification paradigm; (2) automated nonparametric identifiability assessment under observational, experimental, or hybrid data regimes; and (3) an open-source Python package (installable via pip) that outputs estimable closed-form symbolic expressions, thereby enhancing rigor, interpretability, and reproducibility in causal modeling.

This work addresses the challenge of determining causal effect identifiability in linear structural causal models with latent confounding, a problem traditionally hindered by the double-exponential computational complexity of Gröbner basis methods. The authors propose a novel symbolic computation algorithm that, for the first time, decides rational identifiability of causal effects in quasipolynomial time and efficiently computes the lowest-degree identification formula under a given maximum degree constraint. By integrating techniques from algebraic geometry with causal inference theory, the method substantially enhances algorithmic scalability and practical applicability, thereby overcoming a longstanding computational bottleneck in the field.

This work addresses causal reasoning in dynamic environments where objects and their relationships evolve over time by proposing the Relational Structural Causal Model (RSCM), which formalizes, for the first time, the problem of causal identifiability in relational settings. By introducing relational causal graphs and symbolic identification criteria, the framework enables both interventional and counterfactual reasoning and generalizes to novel object configurations involving unobserved confounders. Building upon this foundation, the authors develop a provably correct relational neural causal network. Empirical evaluation in a dynamic traffic simulation—featuring vehicles, traffic signals, and pedestrians—demonstrates that the proposed model significantly outperforms non-relational baseline approaches.

Existing causal identification methods rely heavily on probabilistic semantics, rendering them inapplicable to non-probabilistic causal systems such as databases, hardware description languages, distributed systems, and modern machine learning frameworks. Method: We propose the first purely syntactic causal identification framework grounded in symmetric monoidal categories, fully decoupling the syntactic structure of causal models from their semantic interpretation. By syntactically reconstructing ADMG graph structures, the ID algorithm, and backdoor/front-door adjustments, we eliminate reliance on probabilistic assumptions. Contribution/Results: Our framework enables categorical compositional transformations, yielding a verifiable and programmable general causal identification algorithm. Empirical validation confirms its effectiveness in complex non-probabilistic systems. This work establishes a foundational theoretical basis for formal and automated causal reasoning, advancing beyond probability-centric paradigms toward category-theoretic formalization of causality.

This study addresses the tendency of reasoning models to yield false affirmatives for non-identifiable queries in causal identification, alongside the absence of reliable evaluation metrics. To this end, it proposes CERTID, a formalized pipeline that leverages structural causal models and the ID algorithm to certify identifiability and verify derived expressions. Through theoretical results, the framework mitigates structural leakage, rectifies non-identifiable queries, and establishes a rigorous hierarchical guarantee mechanism. Empirical findings reveal that accuracy is not a reliable proxy for trustworthiness, with spurious claim rates varying up to 17-fold across models. Notably, CERTID achieves 97–100% decision accuracy on novel graphs, effectively resolving the longstanding challenges of evaluation deficiency and scoring difficulty in this domain.

This study addresses the challenge of efficiently selecting observational variables for causal effect identification in partially specified causal models. Focusing on iterative identification within semi-Markovian models, this work leverages causal graph theory to establish necessary and sufficient graphical conditions under which previously unidentifiable queries become identifiable. Furthermore, it designs an efficient localization algorithm to optimize navigation through the model space and streamline the search for optimal observation strategies. By refining underlying identification mechanisms to precisely pinpoint critical observation opportunities, this research significantly enhances both the efficiency and accuracy of constructing identification strategies in causal inference.

研究解决了混合数据集中因果方向识别问题，通过有序logit模型和单参数指数族分布证明了节点间因果方向的可识别性，并提出基于评分的搜索和连续优化框架。

This study addresses the problem that variable clustering operations in causal graphs may compromise the identifiability of causal effects, leading to erroneous inferential conclusions. To resolve this issue, we propose a class of identification-invariant clustering operations grounded in c-component structural analysis, ensuring that the identifiability of causal effects after clustering remains strictly consistent with that of the original graph. By integrating causal inference theory with graph-theoretic analytical methods, this work overcomes the challenge of preserving non-identifiability and achieves complete invariance in causal effect identification under clustering operations. The effectiveness of the proposed approach is further validated through empirical evaluations in practical scenarios.

This study addresses the limitation that performance evaluations of existing causal foundation models are confounded by observational perspectives, lacking a fair comparison benchmark. We propose CausalIDView, a multi-perspective benchmark that enables controlled experiments by fixing the structural causal model and target estimand while varying only the observational perspective. Furthermore, it adopts a modular approach that decouples predictive tabular foundation models from specific identification mechanisms, facilitating equitable comparisons across different identification strategies. Our experiments reveal significant instability in model rankings across perspectives, with no single model consistently outperforming others. Additionally, we demonstrate that this modular approach surpasses several existing causal foundation models in performance, establishing a new paradigm for the robustness evaluation of causal models.