This repository curates research at the intersection of causal inference and natural language processing, providing a valuable resource for researchers and practitioners in both fields. It aims to bridge the gap between understanding language and drawing causal conclusions from it, offering a structured overview of methods, datasets, and applications.

How It Works

The collection is organized by how text is utilized within causal inference frameworks: as treatment, mediator, outcome, or confounder. It also categorizes papers by their application domains, such as social sciences, marketing, and healthcare, and by specific NLP tasks like causal explanations and robustness. This structure allows users to quickly find relevant work based on their specific research questions or methodological interests.

Quick Start & Requirements

This repository is a curated list of research papers and codebases, not a runnable software package. Users can access the listed papers and their associated code repositories directly via the provided links.

Highlighted Details

• Comprehensive categorization of papers by the role of text in causal inference (treatment, mediator, outcome, confounder).
• Extensive coverage of applications across various domains including social sciences, linguistics, marketing, mental health, and law.
• Includes datasets and simulations for both semi-simulated and fully synthetic scenarios.
• Features learning resources and blog posts that provide overviews and methodological insights.

Maintenance & Community

This is a community-driven collection. Pull requests are welcome for adding new papers or codebases. Specific contributors or maintainers are not highlighted in the README.

Licensing & Compatibility

The licensing of individual papers and codebases varies and is not specified centrally. Users must refer to the licenses of each linked resource.

Limitations & Caveats

This repository is a curated list and does not provide a unified software framework or direct execution environment. Users are responsible for accessing and integrating the individual research components.