Integrating Unstructured Text into Causal Inference: Empirical Evidence from Real Data

2026-02-15

AI Summary

This study addresses the limitation of traditional causal inference methods, which rely heavily on structured data that are often missing or incomplete in real-world settings. To overcome this challenge, the authors propose a novel framework based on Transformer language models that systematically extracts causal effects directly from unstructured text. For the first time, they conduct a comprehensive validation by comparing textual causal estimates against those derived from structured data across three levels: population, subgroup, and individual. The results demonstrate that causal inferences drawn from text data are consistent and reliable, substantially expanding the applicability of causal analysis to scenarios where structured data are unavailable. This work establishes a new paradigm for leveraging large-scale textual data in causal research.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Web Mining and Content Analysis: Bridging structured and unstructured data
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Causal inference, a critical tool for informing business decisions, traditionally relies heavily on structured data. However, in many real-world scenarios, such data can be incomplete or unavailable. This paper presents a framework that leverages transformer-based language models to perform causal inference using unstructured text. We demonstrate the effectiveness of our framework by comparing causal estimates derived from unstructured text against those obtained from structured data across population, group, and individual levels. Our findings show consistent results between the two approaches, validating the potential of unstructured text in causal inference tasks. Our approach extends the applicability of causal inference methods to scenarios where only textual data is available, enabling data-driven business decision-making when structured tabular data is scarce.

Problem

• causal inference
• unstructured text
• structured data
• business decisions
• real-world scenarios

Innovation

• causal inference
• unstructured text
• transformer-based language models
• textual data
• business decision-making