<![CDATA[Twitter is popular social media in great demand because it provides information needed by many internet users. Such information can be in the form of opinions, questions or review of a product's good or bad. Diverse smartphone product reviews make it difficult for companies to know people's interests and opinions on the smartphone product. To find a solution for this problem, sentiment analysis system is needed on tweets about smartphone products. This research conducted a sentiment analysis with the K-Nearest Neighbor (textual) method to carry out the classification process and add weighting features to the number of likes (non-textual). The result of combining intellectual and non-textual weighting with certain constants is α constans and β constans will produce a class of positive and negative sentiments. The data was used is taken from Twitter in the form 300 data tweets of smartphone product reviews. The test results of 210 training data and 90 test data with textual weighting obtained an accuracy of 91.01%, using only non-textual weighting of 68.53% and combining textual and non-textual weighting resulted an accuracy of 94.38%]]>
Copyrights © 2019
