Movilizer Application with Genre and Rating Classification Using NW-KNN Method

Abstract

Information about movies can be easily seen in cyberspace. However, not all film sites present relevant and accurate information as examples of high rating films but have bad comments. In addition, there is a review that has not been accompanied by ratings and the genre is unknown. The classification of input data in the form of text will be processed and classified into the same or similar class using the Neighbor-Weighted K-Nearest Neighbor (NW-KNN) method. The NW-KNN method is able to classify well for data that is not evenly distributed by giving weights to each class in the system. The description text of the film will be classified into 10 classes with the number of training data as many as 1028, while the movie review text will be classified into 5 classes with the number of training data as many as 10032. The results of system testing indicate that the NW-KNN method produces an accuracy of 96.6% film genre and 86.85% to classify film reviews into movie ratings.