Perancangan Sistem Analisis Sentimen Komentar Pelanggan Menggunakan Metode Naive Bayes Classifier

Evasaria Magdalena Sipayung, Herastia Maharani, Ivan Zefanya

Abstract


Abstract
Grand Royal Panghegar is a company runs in hospitality sector located in Bandung. This hotel faced problem in getting the meaning or conclusion of comments from customes about hotel’s products and services, because the amount of the comments reached about 675 comments every year. To overcome the problem, developing a tool named sentiment analysis system. This system supports the hotel to get the meaning from the large comments using Naive Bayes Classifier (NBC) method. This method classified the categories that will reviewed by the hotel and divided by positive and negative sentiment, so hotel can be evaluated by the customer satisfaction to products and services that provided computerized, spesific, and systematic. The result from this research is six categories that reviewed with 55 keywords of nouns. From this research got 120 keywords sentiment with 66 of positive sentiments and 54 of negative sentiments. The result of processing from 175 training sets by system can be concluded that the most often sentiment that appear is sentiment positive for 155 comments and 20 comments of negatif sentiment. And then for the highest positive sentiment category is hotel room with 73 comments and 17 comments for the highest negative sentiment category. The accuration of this system to determine the category is 77.14% and the precision to determine the sentiment is 99.12%, recall 72.9%, and accuration is 75.42%.
Keywords: sentiment analysis, comments, naïve bayes classifier,


Abstrak
Hotel XYZ mengalami kesulitan untuk mendapatkan makna atau kesimpulan dari keseluruhan komentar yang diberikan pelanggan terhadap produk dan layanan hotel dikarenakan banyaknya komentar yang ada, pertahun mencapai 675 komentar. Sistem analisis sentiment analysis system bertujuan untuk membantu pihak hotel dalam mendapatkan makna dari komentar yang banyak dengan menggunakan metode Naive Bayes Classifier (NBC). Metode ini mengelompokan komentar berdasarkan kategori-kategori yang ditinjau oleh hotel. Komentar dibagi berdasarkan sentimen positif dan negatif, sehingga dapat dievaluasi kepuasan pelanggan terhadap produk dan jasa yang disediakan secara terkomputerisasi dan spesifik. Hasil dari penelitian yang dilakukan mendapatkan enam kategori yang ditinjau dengan 55 keyword kata benda, terdapat 120 keyword sentimen dengan 66 kata sentimen positif dan 54 kata sentimen negatif. Hasil pengolahan terhadap175 data latih disimpulkan bahwa hasil klasifikasi sentimen yang didapat adalah sentimen positif sebanyak 155 komentar dan sentimen negatif sebanyak 20 komentar. Kategori sentimen positif terbesar adalah kategori kamar sebesar 73 komentar dan kategori dengan sentimen negatif terbesar adalah kategori kamar sebesar 17 komentar. Tingkat akurasi dalam penentuan kategori adalah sebesar 77.14% dan 75.42% dalam penentuan sentimen memiliki tingkat precision 99.12% dan recall 72.9%.
Kata kunci: analisis sentimen, komentar, naive bayes classifier


Full Text:

958-965

Refbacks

  • There are currently no refbacks.


SSN ONLINE      : 2355-4614


Recommended Tools

 

Logo Ithenticate Logo Grammarly Logo ithenticate Logo Mendeley

 

Indexing Sites

 

Logo Grammarly Logo Mendeley Logo Mendeley Logo Mendeley