Back to my old tricks using {openai} to categorise some text.
The claims are the most interesting part of this data. I could have simply summarised the words in the claim, instead, I asked chatGPT to describe the claim in 3 words to get the essence of the claim. This would be different every time you run it (maybe that’s a good test to understand the variation in these language models) but useful for standardising each claim so they only contribute 3 words.
Code 👉 Github
