DEVELOPMENT OF A K-NN MODEL TO PREDICT THE POLARITY OF KOREAN GAME REVIEW COMMENTS

The purpose of this paper is to develop the machine learning model to evaluate game software in quantitative scale using opinion mining with the review comments which are provided by users in Korean. To do this, we first decompose the review comments into a lot of meaningful morphemes, and second construct a dictionary for opinion mining. Third, we develop a k-NN model to predict the polarity of review comment. Finally, we predict the polarity for each review comment which is included in validation data set by the model. The experimental results of the developed model are performed by the model which is implemented by JAVA and R language.