ANALISIS SENTIMEN PADA ULASAN APLIKASI JAKARTATERKINI (JAKI) DI GOOGLE PLAY STORE MENGGUNAKANMETODE SUPPORT VECTOR MACHINE

• Main Author: Daniel Dwi Eryanto Manurung, (Author)
• Format: Book
• Published: 2022-07-08.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Link Metadata
• Notes: http://repository.upnvj.ac.id/19779/1/ABSTRAK.pdf; http://repository.upnvj.ac.id/19779/2/AWAL.pdf; http://repository.upnvj.ac.id/19779/3/BAB%20I.pdf; http://repository.upnvj.ac.id/19779/4/BAB%20II.pdf; http://repository.upnvj.ac.id/19779/5/BAB%20III.pdf; http://repository.upnvj.ac.id/19779/6/BAB%20IV.pdf; http://repository.upnvj.ac.id/19779/7/BAB%20V.pdf; http://repository.upnvj.ac.id/19779/8/DAFTAR%20PUSTAKA.pdf; http://repository.upnvj.ac.id/19779/9/RIWAYAT%20HIDUP.pdf; http://repository.upnvj.ac.id/19779/10/LAMPIRAN.pdf; http://repository.upnvj.ac.id/19779/11/HASIL%20PLAGIARISME.pdf; http://repository.upnvj.ac.id/19779/12/ARTIKEL%20KI.pdf
• Keywords: Q Science (General); QA75 Electronic computers. Computer science; T Technology (General)

Over time, the Jakarta Terkini (JAKI) application continues to develop well. However some users have complained of their dissatisfaction with the services provided by the app which is seen through the comments review. Several types of complaints from JAKI users include vaccine certificates that are not available, vaccine quotas are always full, difficulties in data entry and so on. Based on these problems, this study conducted a sentiment analysis on the JAKI application through commentary reviews to provide information to the public regarding application performance using the Support Vector Machine (SVM) method and the chi-square method for feature selection. The number of datasets obtained is 1000 data and has not been labeled. In classifying the data obtained, it is necessary to label the data and clean the data first before the text processing stage, then the data is given a weight for each word with Term Frequency-Inverse Document Frequency (TF-IDF) which will be used as a feature, then feature selection is carried out with chi -square. The next stage is the distribution of data by 80% 20% and classified by the SVM method. The results obtained from this study are 66 features selected and modeling using the RBF kernel, C = 40 and gamma = 0.1 from SVM, it is found that 120 test data (testing) get 97% accuracy, 100% precision, 93% recall, and a specificity of 100%.