LEARNING TO CLASSIFY TROPICAL DISEASE WEB PAGES FROM LARGE INDONESIAN WEB DOCUMENTS

In the era of internet technology, many cases of tropical diseases like malaria, leprosy, and dengue fever are reported online. These online facts can be very useful to track the spread of the diseases. Studies in classifying tropical disease web pages from a large set of Indonesian web pages have not yet been recognized. In this paper, we built classifiers using Support Vector Machine (SVM), Naive Bayesian, and K-Nearest Neighbors. We generated dictionaries of n-gram terms for both positive (tropical disease) and negative (non-tropical disease) classes and used the dictionaries to extract feature attributes of the pages. The experimental results show that SVM with polynomial kernel is the best classifier model when compared to the other models and methods. The F-measure and accuracy of the model are 95.52% and 99.59% respectively.