Optimization of Police Mobile Brigade Corps Sentiment Analysis Based on Post on Twitter Using Algorithm Support Vector Machine and Naive Bayes with Particle Swarm Optimization
Endah Putri Purnamasari (a), Bryan Pratama (b), Windu Gata (c), Dedi Dwi Saputra (d), Deny Novianti (e), Ahmad Yusuf Malik (f)
a. STMIK NUSA MANDIRI
Jakarta, Indonesia
14002189[at]nusamandiri.ac.id
b. STMIK NUSA MANDIRI
Depok, Indonesia
Bryanp2803[at]gmail.com
c. STMIK NUSA MANDIRI
Jakarta, Indonesia
Windu_gata[at]yahoo.com
d. STMIK NUSA MANDIRI
Jakarta, Indonesia
14002190[at]nusamandiri.ac.id
e. STMIK NUSA MANDIRI
Jakarta, Indonesia
ddenynovianti[at]gmail.com
f. STMIK NUSA MANDIRI
Jakarta, Indonesia
14002182[at]nusamandiri.ac.id
Abstract
Brimob is a special operating unit that is a paramilitary property of the indonesian national police. The Brimob Corps is also known as one of the oldest unit in the polri organization. Currntly, the national police corps brigade is busy being discussed in the ral world and cyberspace, especially on social media twitter. Many opinions about the national police corps brigade so that there are positive and negative opinions. Social media twitter is now one of the places to disseminate information about the national police coprs brigade. In the previous study, the maximum accuracy was still lacking. The case of this study uses text mining techniques with the support vector machine (SVM) and naive bayes (NB) methods with Particle Swarm Optimization (PSO) with the addition of 150 data sheets, NB having an accuracy value of 85.67% with AUC 0.8188 while NB PSO obtaining 89,69 % accuracy with AUC 0.875. SVM has an accuracy value of 93.40% with AUC 0.981, while SVM PSO has accuracy value of 94,85% and AUC 0,978. the best optimization application in this model is the SVM PSO can provide solutions to classification problems in this case sentiment analysis. SVM PSO algotirthm provides a solution for analyzing sentiments from the content of various online media news optimally.
Keywords: brimob; mako brimob; sentiment analys
Topic: International Symposium on Social Sciences, Humanities, Education, and Religious Studies