causaltext/causal-text-papers

Organizes papers across multiple research dimensions—text as treatment/mediator/outcome/confounder in causal models—alongside domain applications spanning social sciences, law, and online discourse. Curates both foundational datasets (semi-simulated and fully synthetic benchmarks) and accessible learning resources including surveys and blog posts. Companion repositories link directly to implementations across TensorFlow and PyTorch, enabling reproducibility of techniques like causal BERT embeddings and propensity-score adjustment methods.

814 stars. No commits in the last 6 months.