Sentiment Analysis of Indonesia’s National Economic Endurance using Fuzzy Ontology-Based Semantic Knowledge

Abstract

Campaigns of election candidates often enliven the social media platforms that they chose. The total number of social media users in Indonesia has reached approximately 130 million users. Making use of this momentum where the social media is very active in the year of campaigns and elections, we attempt to mine the public sentiment in terms of national endurance using fuzzy ontology-based semantic knowledge. A regular ontology is usually considered rather ineffective in extracting information from tweets; thus, we use the concept of fuzzy ontology-based semantic knowledge. Fuzzy ontology-based semantic knowledge is one method of sentiment analysis using combined approach of lexicon-based, ontology-based, and fuzzy logic. This method gives results whether a tweet is categorized as strong negative, negative, neutral, positive, or strong positive. Moreover, a regular ontology is unable to classify a tweet into sentiment categories when that tweet has more than one SentiWord value. From the 2032 tweets with sentiments, we found 205 tweets having more than one SentiWord values. Therefore, the application of FuzzyDL is needed to solve this problem. Using this method, we obtain the accuracy score of 78%, precision score of 93%, recall score of 73%, and F-measure score of 82%.