LLM scale's impact on ontology learning studied across Qwen and GPT models

A new study published on arXiv investigates the impact of Large Language Model (LLM) scale on ontology learning performance. Researchers evaluated 13 models, including variants from the Qwen3.5 and Qwen3.6 lineages, using the OntoLearner pipeline across biomedical and materials science domains. The findings indicate that while increasing parameter count in dense models generally improves precision, the effect of scale is not uniform across all tasks and domains. Notably, dense 27B models outperformed larger sparse models on term typing, and Mixture-of-Experts models showed stronger results in taxonomy discovery. AI

IMPACT Provides empirical guidance for selecting LLMs in ontology engineering, suggesting model size alone is insufficient.

RANK_REASON The cluster contains an academic paper detailing a controlled study on LLM scale for ontology learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• generative pre-trained transformer
• Hamed Babaei Giglou
• Hugging Face
• Materials Data Science
• OntoLearner
• Qwen 3.5
• Qwen3.6

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →