ARIANTO, QUINCY MAYFERTA PRADANI PUTRI (2023) Analisis Sentimen Twitter Menggunakan Metode N-Gram, K-Means Dan Multinomial Naive Bayes (Studi Kasus: BPJS). Undergraduate thesis, UNIVERSITAS PEMBANGUNAN NASIONAL "VETERAN" JAWA TIMUR.
|
Text (Cover)
19081010127-cover.pdf Download (631kB) | Preview |
|
|
Text (Bab 1)
19081010127-bab1.pdf Download (140kB) | Preview |
|
![]() |
Text (Bab 2)
19081010127-bab2.pdf Restricted to Repository staff only until 31 May 2025. Download (296kB) |
|
![]() |
Text (Bab 3)
19081010127-bab3.pdf Restricted to Repository staff only until 31 May 2025. Download (645kB) |
|
![]() |
Text (Bab 4)
19081010127-bab4.pdf Restricted to Repository staff only until 31 May 2025. Download (1MB) |
|
|
Text (Bab 5)
19081010127-bab5.pdf Download (128kB) | Preview |
|
|
Text (Daftar pustaka)
19081010127-daftarpustaka.pdf Download (136kB) | Preview |
|
![]() |
Text (Lampiran)
19081010127-lampiran.pdf Restricted to Repository staff only until 31 May 2025. Download (1MB) |
Abstract
The Social Security Organizing Agency or Badan Penyelenggara Jaminan Sosial (BPJS) is a legal entity established by the Indonesian government to provide social security to its citizens. With the increasing number of BPJS participants, opinions related to the legal entity are increasingly varied. Opinions from BPJS users can be a constructive input for the government to improve service quality according to what is needed by participants. Social media today has an important role in disseminating information as well as a medium for expressing feelings and opinions. The amount of information data on social media attracts the attention of several parties in collecting important data for various purposes such as improving service quality. Opinions related to discussions or products in social media channels are currently dynamic and constructive, which causes the meaning of sentences to become more complex to understand. To make it easier to understand the meaning of an opinion in social media, information extraction is carried out in the form of sentiment analysis. Several studies related to sentiment analysis have been conducted with various methods to improve accuracy in information extraction. Approaches with combined methods related to sentiment analysis research have also been carried out by combining two methods such as k-means with naive bayes, and n-gram with naive bayes. The combination of these methods produces quite good accuracy but is not optimal. The k-means method in combining the k-means and naive bayes methods still experiences several errors in sentiment clustering so that it still requires manual assistance in clustering sentiments. Therefore, this research aims to conduct sentiment analysis with a combination of N-Gram, K-Means, and Multinomial Naive Bayes methods to improve accuracy in understanding sentiment values in text data from social media. The n-gram method can help in clustering data with the k-means method by reducing unimportant words, and optimizing the representation of words to be grouped. Therefore, it is assumed that if the accuracy of clustering with k-means increases, the accuracy of multinomial naive bayes will also increase in classifying sentiment data. The data used in this study amounted to 500 data with the use of data for training data by 75% and test data by 25%. Based on the evaluation assessment, sentiment analysis with the combination of n-grams, k-means, and multinomial naïve bayes produces good accuracy, precision, recall, and F1-score with an accuracy value of 80.80%, precision of 79.38%, recall of 80.80%, and F1-Score of 80.08%.
Item Type: | Thesis (Undergraduate) | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Contributors: |
|
||||||||||||
Subjects: | T Technology > T Technology (General) | ||||||||||||
Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
Depositing User: | Quincy Quincy Arianto | ||||||||||||
Date Deposited: | 06 Jun 2023 01:58 | ||||||||||||
Last Modified: | 06 Jun 2023 01:58 | ||||||||||||
URI: | http://repository.upnjatim.ac.id/id/eprint/14312 |
Actions (login required)
![]() |
View Item |