High precision method for text feature selection based on improved ant colony optimization algorithm

To reflect the overall impact of feature subset on the classification result and remove the noise features, a new high-precision method was proposed for text feature selection based on the improved ant colony optimization algorithm. A mathematical model for feature selection was established to realize the direct correlation between the feature selection process and the classifier classification process. A new model-solving method composed of the optimized feature selection step and the refined feature selection step was designed and the computational complexity in the process of model solving reduced. A new improved ant colony optimization algorithm based on equivalent routes and local search which improved the quality and stability of the problem solution was proposed. The experiment results on the two datasets show the superiority of the proposed method over the current feature selection methods in terms of classification accuracy.