KYBELE

Title: Knowledge Yield from BiodivΕrsity Literature through Large Language Model (LLM) Extraction

Funding Source: ELIXIR

Budget IMBBC: 27,358€

Start / End Date: 2026 - 2026

Web site URL:


Project Progress: 100%

Principal Investigator:

Stelios Ninidakis

Project Members:

Research Directions:

Marine biodiversity

Bioinformatics and biodiversity informatics

Description:

KYBELE aims to develop an AI-powered, FAIR-compliant system to unlock biodiversity knowledge from genomic and ecological literature repositories (PMC, BHL, Zenodo). By harvesting species data, traits and habitats from BOLD and annotating BiodiversityPMC with domain vocabularies, the project creates structured resources for reuse. Generative LLMs, fine-tuned through LoRA and distillation, will enable interactive literature exploration via a containerised chatbox deployed jointly across the ELIXIR and LifeWatch ERIC portfolios. Through dissemination, training and integration with existing infrastructure, KYBELE delivers sustainable services that strengthen ELIXIR’s biodiversity portfolio, support the BFSP Priority Area and empower researchers with innovative tools and datasets to advance scientific discovery.