Think Global, Act Local: Bayesian Causal Discovery with Language Models in Sequential Data

2025-06-19

AI Summary

Causal discovery under streaming batch data and scarce expert knowledge poses challenges for large language models (LMs), which suffer from hallucination, inconsistency, bias, and difficulty integrating global semantic priors with local observations. Method: We propose the first Bayesian causal discovery framework based on Partial Ancestral Graphs (PAGs), featuring an adaptive LM querying mechanism and a joint bias correction strategy to align LM-derived priors with data-driven inference reliably. We further introduce serialized edge selection and noise-robust parameter estimation to enable Bayesian quantification of parameter uncertainty. Contribution/Results: Our method achieves significantly higher structural accuracy than state-of-the-art approaches across diverse benchmarks. It demonstrates strong robustness against LM hallucination and distributional shift, while enabling principled uncertainty quantification and trustworthy integration of linguistic and statistical evidence.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Causal discovery from observational data typically assumes full access to data and availability of domain experts. In practice, data often arrive in batches, and expert knowledge is scarce. Language Models (LMs) offer a surrogate but come with their own issues-hallucinations, inconsistencies, and bias. We present BLANCE (Bayesian LM-Augmented Causal Estimation)-a hybrid Bayesian framework that bridges these gaps by adaptively integrating sequential batch data with LM-derived noisy, expert knowledge while accounting for both data-induced and LM-induced biases. Our proposed representation shift from Directed Acyclic Graph (DAG) to Partial Ancestral Graph (PAG) accommodates ambiguities within a coherent Bayesian framework, allowing grounding the global LM knowledge in local observational data. To guide LM interaction, we use a sequential optimization scheme that adaptively queries the most informative edges. Across varied datasets, BLANCE outperforms prior work in structural accuracy and extends to Bayesian parameter estimation, showing robustness to LM noise.

Problem

• Bridging gaps in causal discovery with sequential batch data
• Integrating noisy LM knowledge with local observational data
• Addressing LM-induced biases and hallucinations in causal estimation

Innovation

• Hybrid Bayesian framework integrating LM and data
• Shift from DAG to PAG for ambiguity handling
• Adaptive edge querying for optimal LM interaction