function to ask LLM to generate part of the ontology

[VladimirAlexiev]

• Use case: there's a standard/narrative (eg DNV's "corrosion under insulation"), and you want to follow it with ontological descriptions.
• Idea: ask LLM to generate part of the ontology corresponding to a selected paragraph/section
• Important: in the prompt, give details about the format to follow (eg org sections, description lists) and annotation properties to include.
  It's best to do that with examples, which constitute n-shot learning.

Libraries/modules:

• https://www.reddit.com/r/emacs/comments/11xii20/why_emacs_is_the_best_platform_for_llms/ claims " there's an awful lot of packages dealing with LLM", and that was a year ago. Here are 2 packages by the same guy:
• https://github.com/ahyatt/llm
• https://github.com/ahyatt/ekg

[johanwk]

This is now straightforward to do. A recent tweak to the format means all OWL axioms/constraints are now written in description lists. This is easier to deal with for typical AI assistants.

I've tried gptel and aidermacs. They work in different ways, but both are quite great.

With the existing generic LLM support, I think it's not really necessary to write functions specifically for ELOT.

However, the instructions in llm-prompt.org need to more work.

[johanwk]

The branch elot-gptel-0.1 has various tools that work with gptel in Emacs. The results are very promising, but it needs more work.