A Chaotic Antlion Optimization Algorithm for Text Feature Selection

Abstract Text classification is one of the important technologies in field text data mining. Feature selection, as a key step processing tasks, used to process high-dimensional feature sets, which directly affects final performance. At present, most widely selection methods academia are calculate importance each for through an evaluation function, and then select subsets that meet quantitative requirements turn. However, ignoring correlation between features effect their mutual combination this way may not guarantee best effect. Therefore, paper proposes chaotic antlion algorithm (CAFSA) solve problem. The main contributions include: (1) Propose (CAA) based on quasi-opposition learning mechanism chaos strategy, compare it with other four algorithms 11 benchmark functions. has achieved higher convergence speed highest optimization accuracy. (2) Study performance CAFSA using CAA when different models, including decision tree, Naive Bayes, SVM classifier. (3) compared eight three Chinese datasets. experimental results show can reduce number improve accuracy classifier, better than methods.