Modeling Climate Obstruction using a RAG Knowledge Graph

The Structure of Climate Obstruction

I've been working in the field of Climate Obstruction recently. A common narrative that I often hear is "oh what a shame that scientists are so bad at communicating complex ideas like climate change". I always thought this was wrong, that the major problem with communicating info about climate change wasn't poor science communication but rather that there are vested interests with very deep pockets who care more about short term profit than the environment. Recently, I discovered that some social scientists have done some research that on this topic. They are called the Climate Social Science Network. They have collected information in various online databases and papers that show how money flows from corporations to think tanks to the media to the public and the end result is to generate confusion and propaganda about climate change so that nothing gets done. They call this Climate Obstruction.

Some resources I encourage people to check out are their Greenwashing model, a free book about Climate Obstruction in Europe, and this very interesting site from the Advertising Standards Association. The figure above is from the Climate Obstruction book. The ASA is an industry organization in the United Kingdom. It turns out that in the UK, you can actually get some serious fines for ads that are false... what a concept! This site from the Columbia Law school is also interesting: Climate Litigation database. As I was bouncing over all those sites I thought "it would be nice to have all this in one place". So I did what I often do these days and asked ChatGPT if there was a way for me to use one of its features to do that. They have some special tools available for a small monthly fee. It said "no but you could create a Retrieval Augmented Generation (RAG) system, would you like me to to explain what RAG is?" And I thought "I know how to build one of those". Thanks to the invaluable help of three undergraduates in data science: Aadarsh Balaji, Jacob Gino, and George Gino, who worked very hard for no money or college credit while they were taking full course loads we put together a RAG system that scraped content from Climate Obstruction web sites into one integrated portal. The document corpus is stored in AllegroGraph and the LLM is accessed via AllegroGraph's magic SPARQL properties. We've had two papers accepted for publication. One in the Formal Ontologies and Information Systems 2025 conference and one in the IEEE Global Humanitarian Technology conference. The FOIS paper is still being polished a bit and it has some of our most interesting results and will be available soon, but here is a preprint of our IEEE paper. All our work is available via our GitHub repository. The documentation for our ontology generated by Widoco, can be found here: Climate Obstruction Ontology documentation.