Traditional and Swarm Intelligent Based Text Feature Selection for Document Classification

In the age of information, the role of text feature selection has become a hot challenge for building automatic document classification models in various application areas. Meanwhile, various powerful schemes dealing with text feature selection are being developed continuously nowadays, but there still exists a research gap for "feature optimization problem": looking for the optimal features by taking account not only local optimism but also global optimism by avoiding the bias problem for unbalance and high dimension text feature. In this paper, we observed swarm-based searching policy which includes Ant Colony Optimization (ACO) and Artificial Bee Colony (ABC) to overcome the local optimization problem of high dimensional feature selection for BBC News document classification. Then, we compared the results using various evaluation measurements for performance and computation cost. The experimental results show that the selected feature using swarm intelligence search is more robust for classification results.