Biomedical named entity recognition using improved green anaconda-assisted Bi-GRU-based hierarchical ResNet model.

Area of Science:

• Biomedical text mining
• Natural Language Processing (NLP)
• Deep Learning (DL)

Background:

• Traditional Named Entity Recognition (NER) methods rely on dictionaries or rules, which have limitations.
• Deep learning (DL) methods offer advancements but struggle with long-distance relationships and require large annotated datasets.
• Biomedical text mining extracts crucial information from scientific literature.

Purpose of the Study:

• To develop a novel deep learning model for improved biomedical named entity recognition (NER).
• To address the challenges of identifying long-distance relationships and data requirements in DL-based NER.
• To enhance the accuracy and efficiency of extracting biomedical entities from text.

Main Methods:

• Proposed the Improved Green Anaconda-assisted Bi-GRU based Hierarchical ResNet BNER model (IGa-BiHR BNERM).
• Utilized the MACCROBAT dataset, pre-processed via Stop Word Filtering, WordNet processing, and tokenization.
• Employed the Robustly Optimized BERT-Whole Word Masking model for feature extraction and word embeddings.
• Applied Improved Green Anaconda Optimization for selecting optimal weight parameters during model training.

Main Results:

• The IGa-BiHR BNERM demonstrated high accuracy in identifying named entities.
• The model achieved a remarkable accuracy rate of 99.11% on the MACCROBAT dataset.
• The pre-processing and feature extraction steps improved the quality and semantic information of the data.

Conclusions:

• The IGa-BiHR BNERM significantly outperforms previous models in biomedical NER.
• The model effectively and accurately identifies biomedical names within text.
• This research represents a substantial advancement in the field of biomedical text mining.