Computational complexity of schema-guided document extraction

This is a topic that interests me deeply! But dang this article needs some supporting documents/visualizations to get its point across better.