Related Experiment Videos
Gene/protein name recognition based on support vector machine using dictionary as features.
Tomohiro Mitsumori1, Sevrani Fation, Masaki Murata
1Graduate School of Information Science, Nara Institute of Science and Technology, Nara, 630-0101, Japan. mitsumor@is.naist.jp
BMC Bioinformatics
|June 18, 2005
Summary
This study presents an automated system for recognizing gene and protein names in biomedical literature using the Support Vector Machine (SVM) algorithm. The system effectively utilizes various features, demonstrating the robustness of SVM for this task.
Area of Science:
- Biomedical Informatics
- Computational Biology
- Natural Language Processing
Background:
- The exponential growth of biomedical literature necessitates automated methods for information extraction.
- Accurate recognition of biomedical named entities, such as genes and proteins, is a foundational step in this process.
Purpose of the Study:
- To develop and evaluate an automated system for gene and protein name recognition.
- To assess the impact of different feature sets and Support Vector Machine (SVM) parameter tuning on recognition performance.
Main Methods:
- Development of an automated recognition system employing the Support Vector Machine (SVM) algorithm.
- Utilization of internal resource features (word, POS, orthography, prefix, suffix, preceding class) and external resource features (dictionary matching).
- Evaluation within Task 1.A of the BioCreAtIvE competition for gene/protein name recognition.
Main Results:
- Dictionary matching features provided a slight improvement in f-score performance.
- The SVM algorithm demonstrated robustness with a high-dimensional feature vector space.
- Individual features showed a marginally positive effect on system performance during SVM learning.
Conclusions:
- Feature selection was not required due to the SVM algorithm's robustness.
- The developed system provides an effective approach for automated gene and protein name recognition in biomedical texts.