Antara sumbar is a news portal aims to meet the public's right to get accurate information and complete information instantly. There are various kind of news presented, one of them is sports news. Rapid improvement on digital information requires prompt news dissemination. As result, web portal needs to be professionally managed. Antara Sumbar is a professionally managed web portal, however there are some shortfall, particularly on sport news categorization that causes difficulties for readers to find the news by its category. The problem can be addressed by BM25 and K-Nearest Neighbor methods. The steps to address the problem are preprocessing document news, calculate the BM25 score of each document news, and classification process using the K-Nearest Neighbor method. The testing process followed 7 k-fold method. Data used on test is 240 training documents and 40 testing documents. Performed test obtaines results in k's value as 20, with precision value = 0,921577, recall values = 0,914286, and f-measure values = 0,917917. The process of classification is affected by the number of document used and k value.
Copyrights © 2019