<![CDATA[Sentiment analysis is a part of text mining, the main focus is to analysis text documents. Sometimes text documents contain non-textual elements, e.g. emojis. Emoji is an Unicode graphic Symbol representation using pictures to express a person's feelings. The algorithm used in this research is Naive Bayes with renewal in addition of non-textual weighting (emoticon). The results of normalised textual and non-textual weightings with Min-Max method will be combined with certain constant values that resulting in both positive and negative sentiments. Data taken from Twitter about 2017 DKI Jakarta elections as much as 900 data tweet. From the accuracy test results, 68,52% were obtained for textual weighting conditions, 74,81% for non-actual weighting, and 73,57% for merging conditions 0,5 for textual and 0,5 for non-textual. From the result of the examination non-textual weighting effect, can be conclude that the non-textual weighting had an effect on the accuracy and classification, with the best multiplier constants when α = 0,4 and β = 0,6 to α = 0,1 and β = 0,9.]]>
Copyrights © 2017
