Text mining of Persian texts based on Cellular Learning Automata and optimizing parameters of SVM

Hossein Alikarami¹ Amirmasoud Bidgoli² Mehdi Sadeghzadeh³

1) Department of Computer, North Tehran Branch, Islamic Azad University, Tehran, Iran. Email:
2) Department of Computer, North Tehran Branch, Islamic Azad University, Tehran, Iran. Email:
3) 3. Department of Computer, Mahshahr Branch, Islamic Azad University, Mahshahr, Iran. Email:

Publication : 4th International Congress on Engineering, Technology and Applied Science - Auckland University of Technology(etas2019.com)
Abstract :
text mining and classification of texts can classify cyberspace information in the minimum time. It can also identify and control the activity field of websites and users in the cyberspace. Words of the text are commonly used as text features in this classification method. Regarding the fact that some selected features are not suitable for classification and lead to an increase in errors in text classification, for the purpose of text mining to decrease the number of features, the optimum features are selected using Cellular Learning Automata, and then these features are given to the multi-layer support vector machine as training data to classify texts. In this way, one of the factors that increases the performance is to optimize the parameters of the SVM. To do this, using the Particle Swarm Optimization (PSO) Algorithm the optimum value for parameters of support vector machine is searched and selected which leads to the minimization of SVM classification error. Training data used in this paper are preprocessed from Hamshahri newspaper, and has been implemented in Matlab. In this way which is one of the supervised learning methods, information is mapped from the current space to another vector space – generally with more dimension – in which linear learning algorithms are applicable. It can do classification with an accuracy of 93.5%.
Keywords : text classification Optimizing SVM parameters Cellular Learning Automata Artificial intelligence.