Gaussian Radial Basis Function Neural Network with Correlation Based Feature Selection Applied to Medical Text Categorization

Öz

Text categorization is an important field for information processing systems. Particularly, medical text processing is a popular research area that makes use of classification algorithms and dimension reduction strategies from machine learning field. In this study, we propose a three stage algorithm to automatically categorize medical text from OHSUMED corpus. In the proposed algorithm, we use Correlation Based Feature Filtering on top of Radial Basis Function Neural Network. The algorithm for 12 sample datasets produces 0.890 in terms macro average F-measure. In this context, both Correlation based Feature Filtering as a feature elimination strategy and Radial Basis Function Neural Network as text categorization algorithm are promising methods

Anahtar Kelimeler

• Machine learning,text categorization,neural networks,feature selection

Kaynakça

1. 1. Pons, A, Gil, P, García, R, Berlanga, L. 2007. Using Typical Testors for Feature Selection in Text Categorization. Lecture Notes in Computer Science, Springer; 643-652.
2. 2. Qirui, Z, Jinghua, T, Huaying, Z, Weiye, T, Kejing, H. Machine Learning Methods for Medical Text Categorization. Circuits, Communications and Systems, Pacific-Asia Conference, 2009, pp 494-497.
3. 3. Yang, Y, Joachims, T. 2008. Text Categorization. Scholarpedia Text Categorization; 4242-4245.
4. 4. Janecek, A, Gansterer, W. On the Relationship Between Feature Selection and Classification Accuracy. JMLR: Workshop and Conference Proceedings, 2009, pp 90-105.
5. 5. Forman, G. 2007. An extensive empirical study of feature selection metrics for text classification. Journal of Machine Learning Resources; 1289-1305.
6. 6. Deng, Z, Tang, S, W, Zhang, M. 2005. An Efficient Text Categorization Algorithm Based on Category Memberships. Fuzzy Systems and Knowledge Discovery; 480-485.
7. 7. Sebastiani, F. 2002. Machine learning in automated text categorization. ACM Computing Surveys; 34: 1-47.
8. 8. Dumais, S. 1998. Using SVMs for Text Categorization. IEEE Intelligent Systems; 13: 21-23.

9. 9. Liao, Y, Vemuri, V, R. Using Text Categorization Techniques for Intrusion Detection. Proceedings of the 11th USENIX Security Symposium, 2002, pp 51-59.
10. 10. Li, Y, H, Jain, A, K. 1998. Classification of Text Documents. The Computer Journal; 41: 537-546.
11. 11. Ozcift, A. 2011. Enhanced Cancer Recognition System Based on Random Forests Feature Elimination Algorithm. Journal of Medical Systems; 1-9.
12. 12. McNamee, P, Mayfield, J. 2004. Character N-Gram Tokenization for European Language Text Retrieval. Information Retrieval; 7: 73-97.
13. 13. Schapire, R, Singer, Y. 2000. BoosTexter: A Boosting-based System for Text Categorization. Machine Learning; 135-168.
14. 14. Mendez, J, Iglesias, E, Riverola, F, Diaz, F, Corchado, J. 2006. Tokenizing, Stemming and Stopword Removal on Anti-spam Filtering Domain. Current Topics in Artificial Intelligence; 449-458.
15. 15. Text-Mining Research Group, University of West Bohemia, Influence of Word Normalization on Text Classification. http://textmining.zcu.cz/publications/inscit20060710.pdf (accessed at 10.01.2018).
16. 16. Lertnattee, V, Theeramunkong, T. 2007. Effects of Term Distributions on Binary Classification. IEICE Transactions on Information and Systems; 1592-1600.
17. 17. Chou, C, Sinha, P, A, Zhao, H. 2010. A Hybrid Attribute Selection Approach for Text Classification. Journal of the Association for Information Systems; 491-518.
18. 18. Hall, M, A, Smith, L, A. Feature subset selection: a correlation based filter approach. Proceedings of the 1997 International Conference on Neural Information, New Zealand, 1997, pp 237-241.
19. 19. Carnegie Mellon University, Pittsburgh. http://boston.lti.cs.cmu.edu/classes/95-65/HW/HW2/ (accessed at 10.02.2018).
20. 20. Dri, A, Abran, A, Mbarki, S. An Experiment on the Design of Radial Basis Function Neural Networks. International Conference on Information & Communication Technologies, 2006, pp 1612-1617.