{"slug": "frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for", "title": "Frontier LLM-based agents can overcome the ontology curation bottleneck for natural phenotypes", "summary": "Frontier large language model-based agents from Anthropic and OpenAI matched the performance of trained human biocurators in linking free-text phenotype descriptions to ontology terms, a task known as phenotype annotation. In a benchmark against a Gold Standard of Entity-Quality annotations, all five agents fell within the range of inter-curator variability of three human experts, while substantially outperforming the Semantic CharaParser NLP tool. The findings suggest that LLM agents can overcome the ontology curation bottleneck that has limited the scaling of cross-study integration of comparative morphological data.", "body_md": "arXiv:2605.28965v1 Announce Type: new\nAbstract: Linking free-text phenotype descriptions to ontology terms, typically referred to as phenotype annotation, is essential for the cross-study integration of comparative morphological data. This labor intensive process has heavily relied on highly trained human experts, which makes it challenging to scale and thus a key bottleneck. Dahdul et al. (2018) established a Gold Standard (GS) of Entity-Quality (EQ) annotations across seven phylogenetic studies and used it to evaluate three human curators and the Semantic CharaParser NLP tool with ontology-based semantic similarity metrics; they reported that machine-human consistency was significantly lower than inter-curator (human-human) consistency. Here we revisit that benchmark with five frontier hosted LLMs from Anthropic and OpenAI, each operating as an \"agentic curator\" within a self-contained workspace that supplies the source publication PDF, the same annotation guide used by the original human curators, the four project ontologies (UBERON, PATO, BSPO, GO), and a validation script. Evaluated against the same Gold Standard, every agent fell within the range of inter-curator variability of the three trained human biocurators of the original study; the best performing agents approached but did not reach the best performing human curator. Agents substantially outperformed Semantic CharaParser on all four metrics.", "url": "https://wpnews.pro/news/frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for", "canonical_source": "https://arxiv.org/abs/2605.28965", "published_at": "2026-05-29 04:00:00+00:00", "updated_at": "2026-05-29 04:20:45.342686+00:00", "lang": "en", "topics": ["large-language-models", "natural-language-processing", "artificial-intelligence", "ai-agents", "ai-research"], "entities": ["Anthropic", "OpenAI", "Semantic CharaParser", "UBERON", "PATO", "BSPO", "GO", "Dahdul et al."], "alternates": {"html": "https://wpnews.pro/news/frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for", "markdown": "https://wpnews.pro/news/frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for.md", "text": "https://wpnews.pro/news/frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for.txt", "jsonld": "https://wpnews.pro/news/frontier-llm-based-agents-can-overcome-the-ontology-curation-bottleneck-for.jsonld"}}