Indonesian Conjunction Rule Based Sentiment Analysis For Service Complaint Regional Water Utility Company Surabaya

Abstract

PDAM (Regional Drinking Water Company) is a company that provides clean water. PDAM develops their services with by considering complaints, suggestions and complaints from users. Over time PDAM services users are increasing thus allowing the number of complaints to also increase and PDAM is impossible to analyze the complaint data using the manual data. In this research proposes ideas to analyze PDAM complaints data with rule based sentiment analysis and categorization methods. The rule based sentiment analysis in this research used twelve rules, where the uniqueness of this rule based is a detection conjunction. Indonesian conjunction detection is the first method available in Indonesia. Detection of conjunction is proposed to find out whether conjunction has an important influence in the meaning of a sentence. The result of sentiment analysis is a score from complaint sentence are negative, positive or neutral. And categorization is a method to provide sentence score including complaints on turbid, leaky water, leakage, meters, usage, or not getting water. An experiment sentiment analysis was conducted on 392 data containing conjunctions and have score manually sentiment. The accuracy value obtained used rule based with conjunctions detection increases 13% than rule based do not use conjunction detection. And the accuracy value of categorization on 100 complaint data are 84% true and 16% false. So for High accuracy values in Conjunction detection needs to notice the context of the sentence and word dictionary and in categorization must also notice to words and write priority categories in the program.