New model of feature selection based chaotic firefly algorithm for arabic text categorization

The dimensionality reduction is a type of problem that appear in the most classification processes. It contains a large number of features; these features may contain unreliable data which may lead the categorization process to unwanted results. Feature selection can be used for reducing dimensionality of datasets and find interesting relevant information. In Arabic language, the number of works applies a meta-heuristic algorithm for feature selection is still limited due to the complex nature of Arabic inflectional and derivational rules as well as its intricate grammatical rules and its rich morphology. This paper proposes a new model for Arabic Feature Selection that combines the chaotic method in the Firefly Algorithm (CFA). The Chaotic Algorithm replaces the attractiveness coefficient in firefly algorithm by the outputs of chaotic application. The enhancement of the new approach involves introducing a novel search strategy which is able to obtain a good ratio between exploitation and exploration abilities of the algorithm. In terms In terms of performance, the experiments of the proposed method are tested using classifiers, namely Naive Bayes (NB), Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) and three evaluation measures, including precision, recall, and F-measure. The experimental findings show that the combining of CFA and SVM classifiers outperforms other combinations in terms of precision.