Perbandingan Kepuasan Pengguna Aplikasi Chatting Berdasarkan Analisa Sentimen dengan Metode Naive Bayes

• Main Authors: Maudina, Agustin (Author), , Endang Wahyu Pamungkas, S.Kom, M.Kom (Author)
• Format: Book
• Published: 2017.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Connect to this object online
• Notes: https://eprints.ums.ac.id/52019/1/Pernyataan%20Publikasi.pdf; https://eprints.ums.ac.id/52019/2/NASKAH%20PUBLIKASI.pdf
• Keywords: H Social Sciences (General); L Education (General); Q Science (General)

The use of social media is beginning develop comprehensive now. One of them, namely the use of the catting application such as BBM, WhatsApp and Line which users can send text messages, picture, videos, voice mail messages, file, even make a free video calls using internet service. Each of these applications has its; own advantage features and its own weaknesses. This research aims to determine the opinion or sentiment of each chatting application by applying naive bayes methods to classify the user's comments to find the highest probabiliy value. Then applied to classify public opinion on the application to the user's response based on positive and negative sentiments. Training data obtained through a series of stages such as data collection, preprocessing, the process of classification itself using naive bayes and compare the apllication chatting. The method of this research will be tested calculating the accuracy of the comparison output system result with manual classification made by human mind. The result of research has been done is the users satisfaction on chat application is more likely on whatsaap with training data 300 opinion sentences provide 208 positive sentiments and 92 negative sentiment, and system performance testing with training data 400 positive opinions and 500 negative opinions using naive bayes term frequency method produces poor accuracy compared chi square feature on naive bayes method.