ontogpt

Why we included this project

OntoGPT is a Python CLI that pulls structured facts out of free text by pairing an LLM with a LinkML schema defining the shape of the data you want. Its SPIRES method does zero-shot extraction, so you don't need hand-labeled examples to get consistent output, and the results are grounded in existing ontologies, meaning entities come back with stable identifiers instead of free-floating strings. That grounding makes the output easy to merge into a knowledge graph or database, which is why teams working on named entity recognition, relation extraction, or knowledge-base construction over biomedical text tend to find it worth a look. The tool runs through LiteLLM, so you can switch between hosted providers or local models without rewriting your extraction setup. Just keep in mind it's a library, not a hosted service; production use means writing some Python glue around it.