<![CDATA[Mobile applications are increasing in use, because many applications are offering convenience for their users. Users who already use the application have the right to review their experiences during application usage. These reviews are useful for new users and application developers. But there are no features in an app store that can classify these reviews into positive or negative categories. These problems can be solved by an automatic process that can analyze the reviews according to positive and negative reviews. The method used for ranking documents is BM25F and as a classification method the Neighbor-Weighted K-Nearest Neighbor (NWKNN) method is used. Testing done using K-fold Cross Validation method to determine the best number of k and confusion matrix for testing each parameter of BM25F and NWKNN. Based on testing conducted on each parameter the BM25F and NWKNN methods can produce a percentage of f-measure and accuracy reaches 97% and 96%. This proves that the NWKNN method can classify the dataset with an unequal number of classes.]]>
Copyrights © 2019
