Comparison of Performance in Text Mining using Categorization of Unstructured Data

Background/Objectives: The text mining would help finding information to the users in the enormous documents. The text mining has been actively developed and utilized in various fields, mainly English-based document, but Study on the Korean text mining has been relatively limited. The importance of the Korean text mining has emerged with increasing big data including Korean text data, the needs for the intensive study and application of Big Data are increasing. Methods/Statistical Analysis: In this study, we compared the performance of these classifications by applying the method of Bayesian methods, k-NN, decision trees, SVM, and as a neural network in classification of unstructured newspaper article into given categories. Findings: In the experiment result, the SVM model has a high F-measure value relative to other models, and has shown stable results in the classification information and recall rate. Also, this model showed a high F-measure value in the classification of a more granular list. Application/Improvements: The methods of k-nn and decision tree show slightly lower performance than SVM, they are turned out to be appropriate models using classification problem cause of having advantages to easy interpretation and short learning time.