<![CDATA[Spam on Instagram (IG) is generally a comment that is considered as irritating because it does not relate to the photos or videos which were commented. Spam on comment section can cause some negative impacts such as making it difficult to follow the discussion on the posted status and making someone’s photo or video looks very popular, commented by a lot of followers despite the fact that most of the comments are actually spam. This research tries to build a model that can identify spam comments on IG. The comment on IG is in text format, so in this research, we use text processing methods. We use Support Vector Machine (SVM) for spam identification. The comment data used in this study were collected from Indonesian actors and artists who are the most followed accounts in IG. We have tested the spam identification model using SVM method resulted in 78.49% of accuracy. This result is better than the baseline model using NB method (77.25%). This research also tested some of the different training data proportions and SVM remains better than NB. Another result of this research are some adaptations needed for preprocessing and stemming stages that must be customized to support Unicode characters and unique symbols that commonly found in IG comments section.]]>
Copyrights © 2017
