Named entity recognition and classification in biomedical text using classifier ensemble.

Area of Science:

• Biomedical Natural Language Processing
• Bioinformatics
• Computational Biology

Background:

• Named Entity Recognition and Classification (NERC) is crucial for extracting information from biomedical texts.
• Biomedical entities such as genes, proteins, DNA, and RNA often have complex structures, posing challenges for recognition.
• Accurate identification of these entities is vital for advancing biomedical research and data analysis.

Purpose of the Study:

• To propose a novel classifier ensemble technique for biomedical NERC.
• To leverage the search capabilities of Genetic Algorithm (GA) for optimizing classifier voting.
• To improve the accuracy and efficiency of extracting complex biomedical named entities.

Main Methods:

• A Single Objective Optimisation approach was developed using Genetic Algorithm (GA).
• GA was employed to quantify class voting across diverse classifiers.
• Multiple models were constructed using Conditional Random Field (CRF) and Support Vector Machine (SVM) with varied feature representations.

Main Results:

• The proposed technique was evaluated on two benchmark datasets: JNLPBA 2004 and GENETAG.
• Achieved an overall F-measure of 75.97% on the JNLPBA 2004 dataset.
• Achieved an overall F-measure of 95.90% on the GENETAG dataset, demonstrating state-of-the-art performance.

Conclusions:

• The proposed GA-based classifier ensemble technique significantly enhances biomedical NERC.
• The system achieves state-of-the-art performance compared to existing methods.
• This approach offers a robust solution for the complex task of identifying biomedical named entities.