Bio Medical Named Entity Recognition Using Large Language Models
M. S. Abishek, Kishore S, S Jayandar, S J Aadith, K. R. Bindu · 2024
The ever-growing volume of biomedical text data necessitates efficient and accurate methods for information extraction. In this context, BioNER serves as a main source that contributes in automating the identification and classification of various biomedicine obj ects like genes, diseases, chemicals and species within the text. The current study aims at investigating the applicability of large language models (LLMs) in solving BioNER tasks. A comparison of the performance of two leading large language models, namely BLOOM and Gemma, in biomedical entity identification and classification is presented. Established benchmark datasets are used to determine which large language model among the two achieves higher accuracy and robustness in BioNER tasks. The evaluation's outcome will be used to incorporate the chosen large language model into an interactive user interface that is easier to understand. The large language model for BioNER applications will be operated and controlled by users through this user interface. To ensure that the biomedical research community can readily accept the use of BioNER methodologies, especially using machine learning methods, it is imperative to provide such a tool. Furthermore, this research can provide faster BioNER task processing and better accuracy if paired with a big language model..