Naive Bayes and Wordcloud for Sentiment Analysis of Halal Tourism in Lombok Island Indonesia

Abstract

Lombok is one of the halal tourist destinations in Indonesia and has been recognized by the world. To examine these assumptions based on sources from tourist opinion, it is necessary to carry out a sentiment analysis whether their presence is as expected. Google Maps is a platform that can show the location of the island of Lombok along with written reviews from tourists who have visited. The collection of review data is done through the Web Scraping technique on the Google Maps Review, then the data is processed using RapidMiner. The algorithm used is Naive Bayes, an algorithm that uses probability or the concept of opportunity in classification. A word cloud visualization is also displayed to bring up words that tourists often say. 1493 data were obtained after Web scraping and cleansing had been labeled with positive and negative sentiment categories. Preprocessing is carried out which includes tokenize, filter token by length, transform case, stopword, and stemming, then classification using the Naive Bayes algorithm. From the results of testing the Naive Bayes algorithm model, an accuracy rate of 74.75%. Word Cloud visualization also found the top words included "indah", "wisata". “pantai”, “alam”, “gunung”, and “masjid”.