A Machine Learning-based Method for Question Type Classification in Biomedical Question Answering
Mourad Sarrouti1, Said Ouatik El Alaoui
1Mourad Sarrouti, Laboratory of Computer Science and Modeling, Sidi Mohammed Ben Abdellah University, Fez, Morocco,
Methods of Information in Medicine
|April 1, 2017
Summary
This study introduces a machine learning approach for classifying biomedical questions into four types: yes/no, factoid, list, and summary. The method achieved 89.40% accuracy, significantly outperforming baseline systems.
Area of Science:
- Biomedical informatics
- Natural Language Processing
Background:
- Biomedical question type classification is crucial for automatic question answering systems.
- Accurate classification determines the appropriate answer extraction algorithm.
Purpose of the Study:
- To develop an automated method for classifying biomedical questions into four categories: yes/no, factoid, list, and summary.
- To enhance the performance of biomedical question answering systems.
Main Methods:
- Utilized machine learning approaches for question classification.
- Extracted features using handcrafted lexico-syntactic patterns.
- Trained classifiers to predict the category of biomedical questions.
Main Results:
- The proposed method achieved a 10-point increase in accuracy over baseline systems.
- Support Vector Machine (SVM) with lexico-syntactic patterns yielded the highest accuracy of 89.40%.
- Demonstrated significant performance improvement on large, annotated BioASQ datasets.
Conclusions:
- The developed method effectively classifies biomedical questions into four predefined categories.
- The approach significantly outperforms existing baseline systems in classification accuracy.
Keywords:
Biomedical question answeringbiomedical informaticsbiomedical question classificationinformation retrievalnatural language processing
