Title: Text Classification Using Computational Model of the Cerebral Cortex

Year of Publication: Nov - 2014
Page Numbers: 9-23
Authors: Koki Hatano , Tomoki Takada and Tomohiro Takagi
Conference Name: The International Conference on Artificial Intelligence and Pattern Recognition (AIPR2014)
- Malaysia

Abstract:


We applied the BESOM cerebral cortex model to text classification. A 20 Newsgroups Data Set is often used as a benchmark data set for text classification and has 20 categories. We used the data set. Although BESOM is currently an imperfect model, we were able to apply it to text classification by using its own basic logic. Results of a text classification task showed that 70.03% of the texts could be classified correctly. In a second task, we reduced the amount of learning data to one text per category (i.e., 20 texts) and found that the precision was 69.79%. Although our proposed method is less accurate than the conventional method using a naïve Bayes classifier, it can perform in acceptable precision and more accurate than naïve Bayes classifier with limited information.