When Does Bigger Help? A Controlled Study of LLM Scale for Ontology Learning

This repository contains the code and data used in our study on the impact of Large Language Model (LLM) scale on ontology learning performance. We present a controlled evaluation of 13 models, including dense and Mixture-of-Experts variants from the Qwen3.5 and Qwen3.6 families as well as proprietary GPT release variants, using the OntoLearner retrieval-augmented generation pipeline. All models are evaluated under the same embedding model, retrieval settings, prompts, decoding parameters, datasets, and metrics across term typing, taxonomy discovery, and non-taxonomic relation extraction tasks on four ontologies from biomedical and materials science and engineering domains. The repository includes evaluation scripts, configuration files, cached ontology data, and result outputs supporting reproducible LLM-assisted ontology engineering experiments.