HEREDITARY latest research takes the stage at CLEF 2026

From 21 to 24 September 2026, the Conference and Labs of the Evaluation Forum (CLEF) 2026 brought together researchers, developers and students working on information access and evaluation in Jena, Germany. Hosted by Friedrich-Schiller-Universität Jena, the event provided a platform for presenting new approaches and discussing advances in information access systems.

HEREDITARY was represented by several project partners, with multiple contributions focusing on the GutBrainIE task, within the BioASQ lab, which addresses the extraction of information about the gut–brain axis from biomedical literature.

Contributions from HEREDITARY partners

Graphwise presented “Do You Need a Frontier LLM, or Can Domain-Specific Encoders Compete in Gut–Brain Information Extraction?”, exploring different approaches to named entity recognition, entity disambiguation and relation extraction. The team achieved second place in the Named Entity Recognition (NER) and Named Entity Recognition and Disambiguation (NERD) subtasks, while its Concept-level Relation Extraction system also ranked second in its respective subtask.

TU Graz also contributed to the GutBrainIE task with “CHASTE: Constrained Holistic Annotator System with onTological Entities”. The work explores a grammar-based approach to entity and relation extraction, combining language models with ontological and natural-language information. A merged run combining the TU Graz and Graphwise approaches achieved a top-precision result in one of the GutBrainIE subtasks.

Aalborg University (AAU) contributed three papers to the GutBrainIE workshop: “GutHub: Tackling the GutBrainIE Task Through Entity Recognition, Disambiguation, and Relation Extraction”, “GetGut: A Multi-Stage Pipeline for Gut-Brain Axis Information Extraction” and “Trusting Gut Instincts: A Fine-Grained Error Analysis of Transformer-Based Structured Data Extraction from Gut-Brain Axis Publications”. Together, the contributions address different aspects of information extraction and the analysis of transformer-based approaches for biomedical literature.

Advancing gut–brain information extraction

The participation of HEREDITARY partners at CLEF 2026 reflects the project’s ongoing work on extracting, structuring and connecting information from biomedical literature, contributing to richer knowledge resources around the gut–brain axis and neurological and neurodegenerative diseases.