Sentiment Analysis and Topic Modelling for Identification of Government Service Satisfaction
| Autor: | R. V. Hari Ginardi, A. Miftah Fajrin, Moh. Nasrul Aziz, Ari Firmanto |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Topic model
Service (business) Information retrieval business.industry Computer science Sentiment analysis 02 engineering and technology Public opinion Latent Dirichlet allocation Support vector machine symbols.namesake Identification (information) ComputingMethodologies_PATTERNRECOGNITION 020204 information systems 0202 electrical engineering electronic engineering information engineering symbols 020201 artificial intelligence & image processing Social media business |
| Zdroj: | 2018 5th International Conference on Information Technology, Computer, and Electrical Engineering (ICITACEE). |
| Popis: | The era of information disclosure and social media tends to make people express their opinions on social media. Indonesia is one of the top five nations social media users in general, especially Twitter. This becomes an interesting thing when trying to see public opinion on a government service. Opinion mining can be used to get information from textual twitter to be processed into an information by classifying existing information into positive information classes and negative information classes. In this research, we try to do opinion mining on public opinion about Identification card (KTP) service in Surabaya city. We compare between supervised and unsupervised methods to see their performance for each classifier. In unsupervised the sentiwordnet approach is used to classify between negative and positive opinions. Supervised Support Vector Machine (SVM) method is used to create a classification model to define an opinion. Before the data is classified, preprocessing steps are used to make the data better. In addition, the Latent Dirichlet Allocation (LDA) approach is used to see for topics that tend to be strong which affects a negative or positive opinion. The result of the classification model by using SVM achieved accuracy rate of 75%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|