Overview of BioNNE Task on Biomedical Nested Named Entity Recognition at BioASQ 2024

Abstract

Recognition of nested named entities, which may contain each other, can enhance the coverage of found named entities. This capability is particularly useful for tasks such as relation extraction, entity linking, and knowledge graph population. This paper presents the organizers’ report on the BioNNE competition, which focused on nested named entity recognition systems in medical texts for both English and Russian. The competition includes three subtasks: Bilingual, English-oriented, and Russian-oriented. Training and validation sets were derived from a subset of the NEREL-BIO dataset, a corpus of PubMed abstracts. For the BioNNE evaluation, eight of the most common medical entity types were selected from the original dataset. Additionally, a novel test set was developed for the shared task, consisting of 154 abstracts in both English and Russian. Held within the framework of the BioASQ workshop, the competition aims to advance research in nested NER within the biomedical domain.