A Two Steps Approach for Afan Oromo Nonfiction Text Categorization

This study presents Afan Oromo text categorizations which use clustering & classification approaches. In natural language such as Afan Oromo, as amount of text documents in electronic format increases, it become difficult to filter, manage, store and process the desired content of information in natural language text. The solution of this problem is developing a tool that categorizes text documents according to their contents. The aim of this study was to design, and implement Afan Oromo nonfiction text categorization model & examining the application of machine learning techniques for automatic Afan Oromo nonfiction text categorization system. Data was collected from Oromia Culture and Tourism Bureau, Oromo cultural center, online electronic documents and other nonfiction books available. In current study, python programming language applied to tokenize, remove stop words and stem Afan Oromo nonfiction text words whereas R programming language was utilized for document indexing, Normalization, cosine similarity, and preparing documents for machine learning. Weka with java is utilized to split Afan Oromo nonfiction text document data set into train set and test set whereas weka tool was utilized for clustering and classification of Afan Oromo nonfiction texts. By using kmean algorithm Afan Oromo nonfiction text document clustering tasks were performed four times to get classes of documents. Among those clustering tasks, one clustering was resulted in cluster with 8 main categories were obtained as good clusters. J48, NaïveBayes, BayesNet, and SMO classifier algorithms were implemented for training text classification model depending on 8 main classes of documents. Among those classifications algorithms, J48 algorithm shows higher performance 94.3755% and hence it was utilized for constructing classification model. From this work it was possible to conclude that machine learning techniques can be applied for Afan Oromo nonfiction text categorization. Further researches also recommend for Afan Oromo nonfiction text Categorization to upgrade the findings.