Text Mining Sentiment Analysis on Social Media Following the Discourse of Converting A Gas Stove to An Electric Stove Ida Bagus Ketut Surya Arnawa (a*), Paula Dewanti (a), Indriyani (a)
a) Institut Teknologi dan Bisnis STIKOM Bali
Jalan Raya Puputan No.86, Dangin Puri Klod, Kec. Denpasar Tim., Kota Denpasar, Bali 80234
*arnawa[at]stikom-bali.ac.id
Abstract
Gas stoves are one of the most frequently used cooking equipment. Gas stoves have several advantages, namely that cooking becomes more practical, easier, and faster. But behind these advantages, there are also disadvantages that gas stoves have. One of the drawbacks is that they pose a risk, such as an explosion of gas cylinders due to inadvertent use. Apart from having the risk of an explosion on a gas stove, in its operation it requires LPG (Liquefied Petroleum Gas) gas, which is one of the products of non-renewable natural resources. Continuous use of LPG gas causes its availability in nature to become increasingly scarce. One of the efforts made by the government is to make a discourse on converting gas stoves to electric stoves. The government^s discourse on converting gas stoves to electric stoves reaps pros and cons for the community. People express their pros and cons through social media. In efforts to analyze sentiment, the help of text mining is needed. By utilizing text mining, it is possible to group the polarities of public opinion. The purpose of grouping public opinion regarding the conversion of gas stoves to electric stoves is to find out the polarity of public opinion, whether it is positive, neutral, or negative. In this study, the TF-IDF and Naive Bayes algorithms are used to maximize the results of sentiment analysis. From the experimental results, the results obtained were a negative sentiment of 64.29% and a positive sentiment of 35.71%. The Naive Bayes algorithm has an accuracy of 80.5%.
Keywords: Naive Bayes Classifier (NBC)- Term Frequency Inverse Document Frequency (TFIDF)- Sentiment Analysis
